diff --git a/ext/MetaData/references/balanced_field.toml b/ext/MetaData/references/balanced_field.toml
index ff504d9c..5b638a26 100644
--- a/ext/MetaData/references/balanced_field.toml
+++ b/ext/MetaData/references/balanced_field.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 200
 parameters = "(t0 = 0.0, m = 79015.8, g = 9.80665, ρ = 1.225, S = 124.7, CD0 = 0.03, AR = 9.45, e = 0.801, CL0 = 0.5, CL_max = 2.0, h_w = 1.0, span = 35.7, α_max = 0.1745, T = 120102.0, r_t0 = 600.0, v_t0 = 70.0, h_t0 = 0.0, γ_t0 = 0.0, h_tf = 10.668, γ_tf = 0.0873, v_min = 0.0, v_max = 100.0, α_min = -0.1745, α_max_ctrl = 0.2618)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 2.4538248057585183e-9
-value = 771.0114003040649
+value = 771.0114003220631
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -24,7 +24,7 @@ value = 771.0114003040649
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = 771.0114003040651
+    objective = 771.011400304065
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,7 +34,7 @@ value = 771.0114003040649
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 771.0114004838803
+    objective = 771.0114004838804
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -54,7 +54,7 @@ value = 771.0114003040649
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = 771.0114003040649
+    objective = 771.0114003040652
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -159,5 +159,46 @@ value = 771.0114003040649
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = 771.0114003040649
+    objective = 771.0114003220631
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=exa solver=madnlp mu_init=1.0e-5"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 54
+        julia = "1.13.3"
+        modeler = "exa"
+        modeling = "OptimalControl"
+        objective = 771.0114003220631
+        os = "Linux"
+        platform = "cron-OptimalControl-robustness-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "madnlp"
+        status = "SOLVE_SUCCEEDED"
+        time = 0.13735
+        violation = 1.0347278589506459e-13
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 1.0e-5
+            print_level = "ERROR"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/beam.toml b/ext/MetaData/references/beam.toml
index 7c400333..6117c329 100644
--- a/ext/MetaData/references/beam.toml
+++ b/ext/MetaData/references/beam.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, tf = 1, x₁_l = 0, x₁_u = 0.1, x₁_t0 = 0, x₂_t0 = 1, x₁_tf = 0, x₂_tf = -1)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
-spread = 6.880267386757687e-8
-value = 8.88928162026725
+spread = 6.880267366774556e-8
+value = 8.889281808581256
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -24,7 +24,7 @@ value = 8.88928162026725
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = 8.889281808507887
+    objective = 8.88928181017181
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,7 +34,7 @@ value = 8.88928162026725
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 8.889282231873594
+    objective = 8.889282231873592
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -44,7 +44,7 @@ value = 8.88928162026725
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = 8.88928184856984
+    objective = 8.889281848569889
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
@@ -54,7 +54,7 @@ value = 8.88928162026725
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = 8.889281808581273
+    objective = 8.88928181130797
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -99,7 +99,7 @@ value = 8.88928162026725
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControl"
-    objective = 8.889281793484152
+    objective = 8.889281850493827
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -144,12 +144,12 @@ value = 8.88928162026725
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = 8.889281793484152
+    objective = 8.889281850493827
     platform = "cron-OptimalControlScalarized-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = 8.889281793484152
+    objective = 8.889281850493827
     platform = "cron-OptimalControlScalarized-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -159,5 +159,48 @@ value = 8.88928162026725
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = 8.88928162026725
+    objective = 8.889281808581256
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=adnlp solver=ipopt mu_init=0.001"
+        cpu = "Intel(R) Xeon(R) Platinum 8370C CPU @ 2.80GHz"
+        iterations = 18
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = 8.889281808581256
+        os = "NT"
+        platform = "cron-OptimalControl-robustness-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "ipopt"
+        status = "first_order"
+        time = 0.119083
+        violation = 1.1102230246251565e-16
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 0.001
+            mu_strategy = "adaptive"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/bioreactor.toml b/ext/MetaData/references/bioreactor.toml
index 36d854d3..fec85304 100644
--- a/ext/MetaData/references/bioreactor.toml
+++ b/ext/MetaData/references/bioreactor.toml
@@ -6,9 +6,9 @@
 agreement = false
 grid_size = 600
 parameters = "(t0 = 0, tf = 200, β = 1, c = 2, γ = 1, halfperiod = 5, Ks = 0.05, μ2m = 0.1, μbar = 1, r = 0.005, y_l = 0, s_l = 0, b_l = 0.001, u_l = 0, u_u = 1, y_t0_l = 0.05, y_t0_u = 0.25, s_t0_l = 0.5, s_t0_u = 5, b_t0_l = 0.5, b_t0_u = 3)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
-spread = 0.000327811483965837
+spread = 0.00032781148396556224
 value = -12.929690242620518
 
     [[trapeze.results]]
@@ -24,7 +24,7 @@ value = -12.929690242620518
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = -12.925452054145852
+    objective = -12.925452054100038
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,7 +34,7 @@ value = -12.929690242620518
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = -12.925451741674866
+    objective = -12.92545174167487
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -44,7 +44,7 @@ value = -12.929690242620518
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = -12.929690242620518
+    objective = -12.929690242620516
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
@@ -54,7 +54,7 @@ value = -12.929690242620518
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = -12.925452054391549
+    objective = -12.925452054463422
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -99,7 +99,7 @@ value = -12.929690242620518
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControl"
-    objective = -12.925451741674909
+    objective = -12.925451741674895
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -146,19 +146,227 @@ value = -12.929690242620518
     configurations = ["JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001"]
     objective = -12.929690242620518
     solves = 4
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001"
+        cpu = "Intel(R) Xeon(R) Platinum 8370C CPU @ 2.80GHz"
+        iterations = 160
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = -12.929690242620518
+        os = "NT"
+        platform = "cron-OptimalControl-robustness-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "ipopt"
+        status = "first_order"
+        time = 3.55974
+        violation = 4.803542869919646e-9
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 0.001
+            mu_strategy = "monotone"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
     [[trapeze.solutions]]
     configurations = ["OptimalControl modeler=exa solver=madnlp mu_init=0.001"]
     objective = -12.929428880081039
     solves = 2
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=exa solver=madnlp mu_init=0.001"
+        cpu = "Intel(R) Xeon(R) Platinum 8370C CPU @ 2.80GHz"
+        iterations = 165
+        julia = "1.13.3"
+        modeler = "exa"
+        modeling = "OptimalControl"
+        objective = -12.929428880081039
+        os = "NT"
+        platform = "cron-OptimalControl-robustness-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "madnlp"
+        status = "SOLVE_SUCCEEDED"
+        time = 19.4614
+        violation = 2.6848551368252593e-9
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 0.001
+            print_level = "ERROR"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
     [[trapeze.solutions]]
     configurations = ["OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5"]
     objective = -12.926202787514399
     solves = 2
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=exa solver=madnlp mu_init=1.0e-5"
+        cpu = "Intel(R) Xeon(R) Platinum 8370C CPU @ 2.80GHz"
+        iterations = 151
+        julia = "1.13.3"
+        modeler = "exa"
+        modeling = "OptimalControl"
+        objective = -12.926202787514399
+        os = "NT"
+        platform = "cron-OptimalControl-robustness-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "madnlp"
+        status = "SOLVE_SUCCEEDED"
+        time = 0.767264
+        violation = 4.057096091372725e-9
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 1.0e-5
+            print_level = "ERROR"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
     [[trapeze.solutions]]
     configurations = ["OptimalControl modeler=exa solver=madnlp barrier=loqo"]
-    objective = -12.925967426083526
+    objective = -12.925946306409168
     solves = 2
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=exa solver=madnlp barrier=loqo"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 129
+        julia = "1.13.3"
+        modeler = "exa"
+        modeling = "OptimalControl"
+        objective = -12.925946306409168
+        os = "Linux"
+        platform = "cron-OptimalControl-robustness-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "madnlp"
+        status = "SOLVE_SUCCEEDED"
+        time = 1.25292
+        violation = 2.71956068775836e-12
+
+            [trapeze.solutions.reference.options]
+            barrier = "loqo"
+            max_iter = 1000
+            max_wall_time = 500.0
+            print_level = "ERROR"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=exa solver=madnlp"]
-    objective = -12.925452054463422
+    objective = -12.925452054150872
     solves = 16
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "636fa809ae3e6c4367343be85d25d63fe119b5bd"
+        configuration = "modeler=adnlp solver=ipopt"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 188
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = -12.925452054150872
+        os = "NT"
+        platform = "cron-OptimalControl-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573"
+        solver = "ipopt"
+        status = "first_order"
+        time = 3.52422
+        violation = 1.1220885332008379e-13
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_strategy = "adaptive"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/brachistochrone.toml b/ext/MetaData/references/brachistochrone.toml
index 0a69f506..295eb702 100644
--- a/ext/MetaData/references/brachistochrone.toml
+++ b/ext/MetaData/references/brachistochrone.toml
@@ -6,10 +6,10 @@
 agreement = false
 grid_size = 500
 parameters = "(g = 9.80665, t0 = 0.0, x0 = 0.0, y0 = 10.0, v0 = 0.0, xf = 10.0, yf = 5.0, u_min = -3.141592653589793, u_max = π)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 0.02249897337664817
-value = 1.802932006652922
+value = 1.802932061276909
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -34,7 +34,7 @@ value = 1.802932006652922
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 1.8029321585265314
+    objective = 1.8029321585265317
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -44,7 +44,7 @@ value = 1.802932006652922
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = 1.8029320612769097
+    objective = 1.802932061276909
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
@@ -54,7 +54,7 @@ value = 1.802932006652922
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = 1.8029320066529235
+    objective = 1.802932006652922
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -149,9 +149,93 @@ value = 1.802932006652922
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = 1.802932006652922
+    objective = 1.802932061276909
     solves = 25
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 22
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = 1.802932061276909
+        os = "Linux"
+        platform = "cron-OptimalControl-robustness-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "ipopt"
+        status = "first_order"
+        time = 0.25709
+        violation = 1.9225225855734607e-9
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 0.001
+            mu_strategy = "monotone"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
     [[trapeze.solutions]]
     configurations = ["OptimalControl modeler=exa solver=madnlp barrier=loqo"]
     objective = 1.843496125847761
     solves = 2
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=exa solver=madnlp barrier=loqo"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 100
+        julia = "1.13.3"
+        modeler = "exa"
+        modeling = "OptimalControl"
+        objective = 1.843496125847761
+        os = "Linux"
+        platform = "cron-OptimalControl-robustness-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "madnlp"
+        status = "SOLVE_SUCCEEDED"
+        time = 5.03583
+        violation = 3.827424155389281e-9
+
+            [trapeze.solutions.reference.options]
+            barrier = "loqo"
+            max_iter = 1000
+            max_wall_time = 500.0
+            print_level = "ERROR"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/bryson_denham.toml b/ext/MetaData/references/bryson_denham.toml
index 7ba434f9..b0c5e118 100644
--- a/ext/MetaData/references/bryson_denham.toml
+++ b/ext/MetaData/references/bryson_denham.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0.0, tf = 1.0, x1_t0 = 0.0, x2_t0 = 1.0, x1_tf = 0.0, x2_tf = -1.0, x1_max = 0.1111111111111111)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 1.4294925612195244e-7
-value = 4.000142925683664
+value = 4.000142996431947
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -24,7 +24,7 @@ value = 4.000142925683664
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = 4.000143422061431
+    objective = 4.000143422061432
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,7 +34,7 @@ value = 4.000142925683664
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 4.000143022023738
+    objective = 4.00014302202381
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -44,7 +44,7 @@ value = 4.000142925683664
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = 4.000142996431951
+    objective = 4.000142996431947
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
@@ -54,7 +54,7 @@ value = 4.000142925683664
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = 4.000143422061428
+    objective = 4.000143422061419
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -99,7 +99,7 @@ value = 4.000142925683664
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControl"
-    objective = 4.000143486269368
+    objective = 4.00014349750112
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -144,12 +144,12 @@ value = 4.000142925683664
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = 4.000143486269368
+    objective = 4.00014349750112
     platform = "cron-OptimalControlScalarized-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = 4.000143486269368
+    objective = 4.00014349750112
     platform = "cron-OptimalControlScalarized-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -159,5 +159,48 @@ value = 4.000142925683664
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = 4.000142925683664
+    objective = 4.000142996431947
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 21
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = 4.000142996431947
+        os = "Linux"
+        platform = "cron-OptimalControl-robustness-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "ipopt"
+        status = "first_order"
+        time = 0.103698
+        violation = 1.1102230246251565e-16
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 0.001
+            mu_strategy = "monotone"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/cart_pendulum.toml b/ext/MetaData/references/cart_pendulum.toml
index 129094ad..c156ab4e 100644
--- a/ext/MetaData/references/cart_pendulum.toml
+++ b/ext/MetaData/references/cart_pendulum.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, g = 9.81, L = 1, m = 1, mcart = 0.5, Fex_l = -5, Fex_u = 5, x_l = -1, x_u = 1, v_l = -2, v_u = 2, tf_l = 0.1, x_t0 = 0, θ_t0 = 0, ω_t0 = 0, θ_tf = π, ω_tf = 0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 1.586433868301789e-6
-value = 1.7439655750225398
+value = 1.7439657987471582
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -24,7 +24,7 @@ value = 1.7439655750225398
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = 1.7439655750325942
+    objective = 1.7439655787788408
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,7 +34,7 @@ value = 1.7439655750225398
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 1.7439660133245751
+    objective = 1.7439667940820989
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -44,7 +44,7 @@ value = 1.7439655750225398
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = 1.7439660141712428
+    objective = 1.7439660142008846
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
@@ -54,7 +54,7 @@ value = 1.7439655750225398
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = 1.7439656616964376
+    objective = 1.7439655750225398
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -99,7 +99,7 @@ value = 1.7439655750225398
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControl"
-    objective = 1.7439667949771815
+    objective = 1.7439667967740533
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -144,5 +144,46 @@ value = 1.7439655750225398
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5"]
-    objective = 1.7439655750225398
+    objective = 1.7439657987471582
     solves = 26
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=exa solver=madnlp barrier=loqo"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 652
+        julia = "1.13.3"
+        modeler = "exa"
+        modeling = "OptimalControl"
+        objective = 1.7439657987471582
+        os = "Linux"
+        platform = "cron-OptimalControl-robustness-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "madnlp"
+        status = "SOLVE_SUCCEEDED"
+        time = 4.65478
+        violation = 3.077650634342177e-11
+
+            [trapeze.solutions.reference.options]
+            barrier = "loqo"
+            max_iter = 1000
+            max_wall_time = 500.0
+            print_level = "ERROR"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/chain.toml b/ext/MetaData/references/chain.toml
index 7d1dfd4a..8ff96b1b 100644
--- a/ext/MetaData/references/chain.toml
+++ b/ext/MetaData/references/chain.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, tf = 1, L = 4, a = 1, b = 3, x₂_t0 = 0, x₃_t0 = 0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 5.178113741056605e-13
-value = 5.0685777900242135
+value = 5.068577790026832
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -24,7 +24,7 @@ value = 5.0685777900242135
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = 5.0685777900268345
+    objective = 5.068577790026836
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,7 +34,7 @@ value = 5.0685777900242135
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 5.0685777900268345
+    objective = 5.068577790026836
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -44,7 +44,7 @@ value = 5.0685777900242135
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = 5.0685777900268345
+    objective = 5.068577790026836
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
@@ -54,7 +54,7 @@ value = 5.0685777900242135
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = 5.068577790026838
+    objective = 5.0685777900268345
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -159,5 +159,48 @@ value = 5.0685777900242135
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = 5.0685777900242135
+    objective = 5.068577790026832
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=adnlp solver=ipopt mu_init=0.001"
+        cpu = "Intel(R) Xeon(R) Platinum 8370C CPU @ 2.80GHz"
+        iterations = 14
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = 5.068577790026832
+        os = "NT"
+        platform = "cron-OptimalControl-robustness-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "ipopt"
+        status = "first_order"
+        time = 0.202702
+        violation = 1.8990808925423153e-11
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 0.001
+            mu_strategy = "adaptive"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/dielectrophoretic_particle.toml b/ext/MetaData/references/dielectrophoretic_particle.toml
index 72c7f211..74b74894 100644
--- a/ext/MetaData/references/dielectrophoretic_particle.toml
+++ b/ext/MetaData/references/dielectrophoretic_particle.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, x_t0 = 1, y_t0 = 0, x_tf = 2, α = -0.75, c = 1, u_l = -1, u_u = 1, tf_l = 0.1)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
-spread = 1.221909871988508e-7
-value = 7.811323680996133
+spread = 1.2219098753996266e-7
+value = 7.811324025977027
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -24,7 +24,7 @@ value = 7.811323680996133
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = 7.811323680996134
+    objective = 7.811323680996133
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,7 +34,7 @@ value = 7.811323680996133
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 7.811324635469484
+    objective = 7.811324635469488
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -159,5 +159,48 @@ value = 7.811323680996133
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = 7.811323680996133
+    objective = 7.811324025977027
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 14
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = 7.811324025977027
+        os = "Linux"
+        platform = "cron-OptimalControl-robustness-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "ipopt"
+        status = "first_order"
+        time = 0.133316
+        violation = 3.994393704687127e-11
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 0.001
+            mu_strategy = "monotone"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/double_oscillator.toml b/ext/MetaData/references/double_oscillator.toml
index a660967f..3239c183 100644
--- a/ext/MetaData/references/double_oscillator.toml
+++ b/ext/MetaData/references/double_oscillator.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, tf = 6.283185307179586, m1 = 100, m2 = 2, c = 0.5, k1 = 100, k2 = 3, u_l = -1, u_u = 1, x₁_t0 = 0, x₂_t0 = 0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 6.248912408876316e-12
-value = 0.0009109230332461941
+value = 0.0009109230332461957
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -24,7 +24,7 @@ value = 0.0009109230332461941
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = 0.0009109230332518861
+    objective = 0.0009109230332518862
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,7 +34,7 @@ value = 0.0009109230332461941
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 0.0009109230332461992
+    objective = 0.0009109230332461991
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -54,7 +54,7 @@ value = 0.0009109230332461941
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = 0.0009109230332518864
+    objective = 0.0009109230332518861
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -159,5 +159,46 @@ value = 0.0009109230332461941
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = 0.0009109230332461941
+    objective = 0.0009109230332461957
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=exa solver=madnlp mu_init=1.0e-5"
+        cpu = "Intel(R) Xeon(R) Platinum 8370C CPU @ 2.80GHz"
+        iterations = 5
+        julia = "1.13.3"
+        modeler = "exa"
+        modeling = "OptimalControl"
+        objective = 0.0009109230332461957
+        os = "NT"
+        platform = "cron-OptimalControl-robustness-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "madnlp"
+        status = "SOLVE_SUCCEEDED"
+        time = 0.028062
+        violation = 5.6107462426124854e-18
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 1.0e-5
+            print_level = "ERROR"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/ducted_fan.toml b/ext/MetaData/references/ducted_fan.toml
index 8526ccdc..1d1d06f4 100644
--- a/ext/MetaData/references/ducted_fan.toml
+++ b/ext/MetaData/references/ducted_fan.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 250
 parameters = "(t0 = 0, r = 0.2, J = 0.05, m = 2.2, mg = 4, μ = 1000, α_l = -0.5235987755982988, α_u = 0.5235987755982988, u₁_l = -5, u₁_u = 5, u₂_l = 0, u₂_u = 17, tf_l = 0.1, x₁_t0 = 0, v₁_t0 = 0, x₂_t0 = 0, v₂_t0 = 0, α_t0 = 0, vα_t0 = 0, x₁_tf = 1, v₁_tf = 0, x₂_tf = 0, v₂_tf = 0, α_tf = 0, vα_tf = 0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
-spread = 7.661904466495749e-8
-value = 1831.8277178167814
+spread = 7.661888603456097e-8
+value = 1831.8278581674542
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -24,7 +24,7 @@ value = 1831.8277178167814
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = 1831.8277178168435
+    objective = 1831.8278488713581
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,7 +34,7 @@ value = 1831.8277178167814
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 1831.8278581696711
+    objective = 1831.8278581693805
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -44,7 +44,7 @@ value = 1831.8277178167814
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = 1831.8277212767207
+    objective = 1831.8277212767446
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
@@ -54,7 +54,7 @@ value = 1831.8277178167814
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = 1831.8277178168887
+    objective = 1831.8278488312462
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -99,7 +99,7 @@ value = 1831.8277178167814
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControl"
-    objective = 1831.8278581674542
+    objective = 1831.827858167453
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -144,12 +144,12 @@ value = 1831.8277178167814
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = 1831.8278581674542
+    objective = 1831.827858167453
     platform = "cron-OptimalControlScalarized-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = 1831.8278581674542
+    objective = 1831.827858167453
     platform = "cron-OptimalControlScalarized-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -159,5 +159,45 @@ value = 1831.8277178167814
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = 1831.8277178167814
+    objective = 1831.8278581674542
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "636fa809ae3e6c4367343be85d25d63fe119b5bd"
+        configuration = "modeler=exa solver=madnlp"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 67
+        julia = "1.13.3"
+        modeler = "exa"
+        modeling = "OptimalControl"
+        objective = 1831.8278581674542
+        os = "NT"
+        platform = "cron-OptimalControl-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573"
+        solver = "madnlp"
+        status = "SOLVE_SUCCEEDED"
+        time = 0.367174
+        violation = 3.0912772341906702e-15
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            print_level = "ERROR"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/electric_vehicle.toml b/ext/MetaData/references/electric_vehicle.toml
index 08173ef4..62b612b9 100644
--- a/ext/MetaData/references/electric_vehicle.toml
+++ b/ext/MetaData/references/electric_vehicle.toml
@@ -6,7 +6,7 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, tf = 1, b1 = 1.0, b2 = 1.0, h0 = 0.1, h1 = 1, h2 = 0.001, α0 = 3, α1 = 0.4, α2 = -1, α3 = 0.1, x_t0 = 0, v_t0 = 0, x_tf = 10, v_tf = 0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 9.253370009051573e-16
 value = 1228.599284481313
@@ -54,7 +54,7 @@ value = 1228.599284481313
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = 1228.599284481313
+    objective = 1228.5992844813134
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -161,3 +161,46 @@ value = 1228.599284481313
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
     objective = 1228.599284481313
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=adnlp solver=ipopt mu_init=0.001"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 4
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = 1228.599284481313
+        os = "Linux"
+        platform = "cron-OptimalControl-robustness-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "ipopt"
+        status = "first_order"
+        time = 0.119365
+        violation = 1.7763568394002505e-15
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 0.001
+            mu_strategy = "adaptive"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/glider.toml b/ext/MetaData/references/glider.toml
index 7492dd77..791dc9f5 100644
--- a/ext/MetaData/references/glider.toml
+++ b/ext/MetaData/references/glider.toml
@@ -6,55 +6,55 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, x_t0 = 0, y_t0 = 1000, y_tf = 900, vx_t0 = 13.23, vx_tf = 13.23, vy_t0 = -1.288, vy_tf = -1.288, u_c = 2.5, r_t0 = 100, m = 100, g = 9.81, c0 = 0.034, c1 = 0.069662, S = 14, ρ = 1.13, cL_min = 0, cL_max = 1.4, tf_l = 0, x_l = 0, vx_l = 0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
-spread = 1.0147162903589398e-10
-value = -1247.9784300226472
+spread = 1.0235125970780203e-10
+value = -1247.9784300208023
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
     modeling = "JuMP"
-    objective = -1247.9784300215604
+    objective = -1247.978430023745
     platform = "cron-JuMP-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt"
     modeling = "JuMP"
-    objective = -1247.9784300216338
+    objective = -1247.9784300226427
     platform = "cron-JuMP-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = -1247.9784300211415
+    objective = -1247.978430023745
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = -1247.9784300216338
+    objective = -1247.9784300226427
     platform = "cron-JuMP-robustness-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = -1247.9784298962754
+    objective = -1247.978429896315
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = -1247.9784298960192
+    objective = -1247.97842989603
     platform = "cron-JuMP-robustness-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = -1247.978429979796
+    objective = -1247.9784299798018
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = -1247.9784299798075
+    objective = -1247.978429979803
     platform = "cron-JuMP-robustness-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = -1247.9784300208023
+    objective = -1247.9784300209178
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -99,7 +99,7 @@ value = -1247.9784300226472
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControl"
-    objective = -1247.9784298963027
+    objective = -1247.9784298962459
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -144,12 +144,12 @@ value = -1247.9784300226472
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = -1247.9784298963027
+    objective = -1247.9784298962459
     platform = "cron-OptimalControlScalarized-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = -1247.9784298962022
+    objective = -1247.9784298962459
     platform = "cron-OptimalControlScalarized-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -159,5 +159,48 @@ value = -1247.9784300226472
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = -1247.9784300226472
+    objective = -1247.9784300208023
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=adnlp solver=ipopt mu_init=0.001"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 38
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = -1247.9784300208023
+        os = "Linux"
+        platform = "cron-OptimalControl-robustness-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "ipopt"
+        status = "first_order"
+        time = 26.2624
+        violation = 4.547473508864641e-13
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 0.001
+            mu_strategy = "adaptive"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/insurance.toml b/ext/MetaData/references/insurance.toml
index 936969df..7176952e 100644
--- a/ext/MetaData/references/insurance.toml
+++ b/ext/MetaData/references/insurance.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, tf = 10, γ = 0.2, λ = 0.25, h0 = 1.5, w = 1, s = 10, k = 0, σ = 0, α = 4, I_l = 0, I_u = 1.5, m_l = 0, m_u = 1.5, h_l = 0, h_u = 25, R_l = 0, H_l = 0, U_l = 0, dUdR_l = 0.001, P_l = 0, I_t0 = 0, m_t0 = 0.001, x₃_t0 = 0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 3.437310685571518e-6
-value = -2.0592143337375415
+value = -2.059213403165573
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -24,7 +24,7 @@ value = -2.0592143337375415
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = -2.0592143322188674
+    objective = -2.0592143337375415
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,7 +34,7 @@ value = -2.0592143337375415
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = -2.059213403165573
+    objective = -2.0592134031655718
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -44,7 +44,7 @@ value = -2.0592143337375415
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = -2.059213992882103
+    objective = -2.0592072555781122
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
@@ -54,7 +54,7 @@ value = -2.0592143337375415
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = -2.059214333311406
+    objective = -2.0592143333119894
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -154,5 +154,41 @@ value = -2.0592143337375415
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = -2.0592143337375415
+    objective = -2.059213403165573
     solves = 28
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "solver=ipopt mu_strategy=monotone"
+        cpu = "AMD EPYC 9V45 96-Core Processor"
+        iterations = 145
+        julia = "1.13.3"
+        modeler = ""
+        modeling = "JuMP"
+        objective = -2.059213403165573
+        os = "NT"
+        platform = "cron-JuMP-robustness-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "ipopt"
+        status = "LOCALLY_SOLVED"
+        time = 1.172
+        violation = 3.9968028886505635e-15
+
+            [trapeze.solutions.reference.options]
+            linear_solver = "mumps"
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_strategy = "monotone"
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            CTBase = "0.30.13"
+            CTModels = "0.19.8"
+            Ipopt = "1.16.0"
+            Ipopt_jll = "300.1400.1902+0"
+            JuMP = "1.32.1"
+            MUMPS_seq_jll = "500.900.100+0"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/jackson.toml b/ext/MetaData/references/jackson.toml
index 6b8f9b2e..00f2bd6f 100644
--- a/ext/MetaData/references/jackson.toml
+++ b/ext/MetaData/references/jackson.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, tf = 4, k1 = 1, k2 = 10, k3 = 1, a_l = 0, a_u = 1.1, b_l = 0, b_u = 1.1, x₃_l = 0, x₃_u = 1.1, u_l = 0, u_u = 1, a_t0 = 1, b_t0 = 0, x₃_t0 = 0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
-spread = 1.4915580303995847e-6
-value = -0.19182194736955152
+spread = 1.4915579296923815e-6
+value = -0.19182194736953223
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -24,7 +24,7 @@ value = -0.19182194736955152
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = -0.19182194736955152
+    objective = -0.19182194736952496
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,7 +34,7 @@ value = -0.19182194736955152
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = -0.19182166125598551
+    objective = -0.19182166125598554
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -54,7 +54,7 @@ value = -0.19182194736955152
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = -0.19182194736953223
+    objective = -0.19182194736952915
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -99,7 +99,7 @@ value = -0.19182194736955152
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControl"
-    objective = -0.1918216612568282
+    objective = -0.1918216612568496
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -144,12 +144,12 @@ value = -0.19182194736955152
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = -0.1918216612568282
+    objective = -0.1918216612568496
     platform = "cron-OptimalControlScalarized-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = -0.1918216612568282
+    objective = -0.1918216612568496
     platform = "cron-OptimalControlScalarized-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -159,5 +159,47 @@ value = -0.19182194736955152
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = -0.19182194736955152
+    objective = -0.19182194736953223
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "636fa809ae3e6c4367343be85d25d63fe119b5bd"
+        configuration = "modeler=adnlp solver=ipopt"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 58
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = -0.19182194736953223
+        os = "NT"
+        platform = "cron-OptimalControl-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573"
+        solver = "ipopt"
+        status = "first_order"
+        time = 0.487541
+        violation = 1.4210410625992154e-11
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_strategy = "adaptive"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/moonlander.toml b/ext/MetaData/references/moonlander.toml
index c81705be..b9c7db2e 100644
--- a/ext/MetaData/references/moonlander.toml
+++ b/ext/MetaData/references/moonlander.toml
@@ -6,10 +6,10 @@
 agreement = false
 grid_size = 500
 parameters = "(t0 = 0, m = 1, g = 9.81, I = 0.1, D = 1, max_thrust = 19.62, tf_l = 0.1, tf_u = 1.0, F₁_l = 0, F₂_l = 0, p₁_t0 = 0, p₂_t0 = 0, dp₁_t0 = 0, dp₂_t0 = 0, θ_t0 = 0, dθ_t0 = 0, p₁_tf = 5, p₂_tf = 5, dp₁_tf = 0, dp₂_tf = 0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 3.505103019045729e-5
-value = 0.9618100531194832
+value = 0.9618109334441546
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -24,7 +24,7 @@ value = 0.9618100531194832
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = 0.9618100531341298
+    objective = 0.9618100542096398
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,17 +34,12 @@ value = 0.9618100531194832
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = 0.961810933422959
-    platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
-    [[trapeze.results]]
-    configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
-    modeling = "JuMP"
     objective = 0.9618109334441546
     platform = "cron-JuMP-robustness-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = 0.9618100664989868
+    objective = 0.9618100531194832
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -84,7 +79,7 @@ value = 0.9618100531194832
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControl"
-    objective = 0.9618109334084259
+    objective = 0.9618109334084258
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -129,9 +124,88 @@ value = 0.9618100531194832
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5"]
-    objective = 0.9618100531194832
-    solves = 22
+    objective = 0.9618109334441546
+    solves = 21
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
+        cpu = "AMD EPYC 9V45 96-Core Processor"
+        iterations = 66
+        julia = "1.13.3"
+        modeler = ""
+        modeling = "JuMP"
+        objective = 0.9618109334441546
+        os = "NT"
+        platform = "cron-JuMP-robustness-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "ipopt"
+        status = "LOCALLY_SOLVED"
+        time = 1.6081
+        violation = 3.552713678800501e-15
+
+            [trapeze.solutions.reference.options]
+            linear_solver = "mumps"
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 0.001
+            mu_strategy = "monotone"
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            CTBase = "0.30.13"
+            CTModels = "0.19.8"
+            Ipopt = "1.16.0"
+            Ipopt_jll = "300.1400.1902+0"
+            JuMP = "1.32.1"
+            MUMPS_seq_jll = "500.900.100+0"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
     [[trapeze.solutions]]
     configurations = ["OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone"]
     objective = 0.9618437655526926
     solves = 1
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=adnlp solver=ipopt mu_strategy=monotone"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 158
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = 0.9618437655526926
+        os = "Linux"
+        platform = "cron-OptimalControl-robustness-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "ipopt"
+        status = "acceptable"
+        time = 22.5685
+        violation = 2.4940405296547397e-9
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_strategy = "monotone"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/mountain_car.toml b/ext/MetaData/references/mountain_car.toml
index 7d524bd3..e75813a1 100644
--- a/ext/MetaData/references/mountain_car.toml
+++ b/ext/MetaData/references/mountain_car.toml
@@ -6,10 +6,10 @@
 agreement = false
 grid_size = 200
 parameters = "(t0 = 0.0, tf_start = 110.0, tf_min = 0.1, pos_t0 = -0.5, vel_t0 = 0.0, pos_tf = 0.5, vel_tf_min = 0.0, pos_min = -1.2, pos_max = 0.5, vel_min = -0.07, vel_max = 0.07, u_min = -1.0, u_max = 1.0, a = 0.001, b = 0.0025, c = 3.0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 0.00016189972119547122
-value = 102.38099470172985
+value = 102.3809949692475
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -24,7 +24,7 @@ value = 102.38099470172985
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = 102.39757015622571
+    objective = 102.39757015622777
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,7 +34,7 @@ value = 102.38099470172985
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 102.3809951930234
+    objective = 102.38099519302341
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -44,7 +44,7 @@ value = 102.38099470172985
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = 102.3809948643563
+    objective = 102.38099486435627
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
@@ -54,7 +54,7 @@ value = 102.38099470172985
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = 102.39757015622693
+    objective = 102.39757015622659
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -99,7 +99,7 @@ value = 102.38099470172985
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControl"
-    objective = 102.38099519301883
+    objective = 102.38099519301736
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -144,12 +144,12 @@ value = 102.38099470172985
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = 102.38099519301883
+    objective = 102.38099519301736
     platform = "cron-OptimalControlScalarized-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = 102.38099519301883
+    objective = 102.38099519301736
     platform = "cron-OptimalControlScalarized-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -159,9 +159,91 @@ value = 102.38099470172985
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = 102.38099470172985
+    objective = 102.3809949692475
     solves = 18
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=exa solver=madnlp barrier=loqo"
+        cpu = "Intel(R) Xeon(R) Platinum 8370C CPU @ 2.80GHz"
+        iterations = 39
+        julia = "1.13.3"
+        modeler = "exa"
+        modeling = "OptimalControl"
+        objective = 102.3809949692475
+        os = "NT"
+        platform = "cron-OptimalControl-robustness-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "madnlp"
+        status = "SOLVE_SUCCEEDED"
+        time = 0.632883
+        violation = 2.09576279941448e-16
+
+            [trapeze.solutions.reference.options]
+            barrier = "loqo"
+            max_iter = 1000
+            max_wall_time = 500.0
+            print_level = "ERROR"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5"]
     objective = 102.39756996372448
     solves = 11
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=exa solver=madnlp mu_init=1.0e-5"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 84
+        julia = "1.13.3"
+        modeler = "exa"
+        modeling = "OptimalControl"
+        objective = 102.39756996372448
+        os = "Linux"
+        platform = "cron-OptimalControl-robustness-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "madnlp"
+        status = "SOLVE_SUCCEEDED"
+        time = 0.0801998
+        violation = 5.48172618408671e-16
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 1.0e-5
+            print_level = "ERROR"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/robbins.toml b/ext/MetaData/references/robbins.toml
index 628cc8bc..1fd67136 100644
--- a/ext/MetaData/references/robbins.toml
+++ b/ext/MetaData/references/robbins.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, tf = 10, α = 3, β = 0, γ = 0.5, x₁_l = 0, x₁_t0 = 1, x₂_t0 = -2, x₃_t0 = 0, x₁_tf = 0, x₂_tf = 0, x₃_tf = 0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 8.340606134238261e-8
-value = 19.46663273582843
+value = 19.46663292374462
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -34,7 +34,7 @@ value = 19.46663273582843
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 19.46663332841832
+    objective = 19.466633328418318
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -54,7 +54,7 @@ value = 19.46663273582843
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = 19.46663316312533
+    objective = 19.466633163125316
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -99,7 +99,7 @@ value = 19.46663273582843
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControl"
-    objective = 19.466633278128064
+    objective = 19.466633278184744
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -144,12 +144,12 @@ value = 19.46663273582843
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = 19.466633278128064
+    objective = 19.466633278184744
     platform = "cron-OptimalControlScalarized-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = 19.466633278128064
+    objective = 19.466633278184744
     platform = "cron-OptimalControlScalarized-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -159,5 +159,48 @@ value = 19.46663273582843
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = 19.46663273582843
+    objective = 19.46663292374462
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 21
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = 19.46663292374462
+        os = "Linux"
+        platform = "cron-OptimalControl-robustness-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "ipopt"
+        status = "first_order"
+        time = 0.130021
+        violation = 4.440892098500626e-16
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 0.001
+            mu_strategy = "monotone"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/robot.toml b/ext/MetaData/references/robot.toml
index b9159381..ebc9e599 100644
--- a/ext/MetaData/references/robot.toml
+++ b/ext/MetaData/references/robot.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 250
 parameters = "(t0 = 0, L = 5, uρ_l = -1, uρ_u = 1, uθ_l = -1, uθ_u = 1, uϕ_l = -1, uϕ_u = 1, ρ_t0 = 4.5, θ_t0 = 0, ϕ_t0 = 0.7853981633974483, dρ_t0 = 0, dθ_t0 = 0, dϕ_t0 = 0, ρ_tf = 4.5, θ_tf = 2.0943951023931953, ϕ_tf = 0.7853981633974483, dρ_tf = 0, dθ_tf = 0, dϕ_tf = 0, ρ_l = 0, θ_l = -3.141592653589793, θ_u = π, ϕ_l = 0, ϕ_u = π, tf_l = 0.1)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 1.9739063663666287e-7
-value = 9.141125184999305
+value = 9.141126989371825
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -34,7 +34,7 @@ value = 9.141125184999305
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 9.141126989371822
+    objective = 9.141126989371823
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -159,5 +159,45 @@ value = 9.141125184999305
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = 9.141125184999305
+    objective = 9.141126989371825
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "636fa809ae3e6c4367343be85d25d63fe119b5bd"
+        configuration = "modeler=exa solver=madnlp"
+        cpu = "AMD EPYC 9V74 80-Core Processor"
+        iterations = 19
+        julia = "1.13.3"
+        modeler = "exa"
+        modeling = "OptimalControl"
+        objective = 9.141126989371825
+        os = "Linux"
+        platform = "cron-OptimalControl-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573"
+        solver = "madnlp"
+        status = "SOLVE_SUCCEEDED"
+        time = 0.0655938
+        violation = 8.595554823465079e-16
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            print_level = "ERROR"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/rocket.toml b/ext/MetaData/references/rocket.toml
index f12bef2c..417a4684 100644
--- a/ext/MetaData/references/rocket.toml
+++ b/ext/MetaData/references/rocket.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, h_t0 = 1, v_t0 = 0, m_t0 = 1, g0 = 1, Tc = 3.5, hc = 500, vc = 620, mc = 0.6, T_l = 0, tf_l = 0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 1.7106896882633451e-6
-value = -1.0128367166830146
+value = -1.0128349840336874
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -44,7 +44,7 @@ value = -1.0128367166830146
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = -1.012836092453478
+    objective = -1.0128360924534778
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
@@ -54,7 +54,7 @@ value = -1.0128367166830146
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = -1.0128367166830134
+    objective = -1.0128367166830141
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -159,5 +159,45 @@ value = -1.0128367166830146
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = -1.0128367166830146
+    objective = -1.0128349840336874
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "636fa809ae3e6c4367343be85d25d63fe119b5bd"
+        configuration = "modeler=exa solver=madnlp"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 23
+        julia = "1.13.3"
+        modeler = "exa"
+        modeling = "OptimalControl"
+        objective = -1.0128349840336874
+        os = "NT"
+        platform = "cron-OptimalControl-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573"
+        solver = "madnlp"
+        status = "SOLVE_SUCCEEDED"
+        time = 0.220189
+        violation = 2.4907888794035993e-11
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            print_level = "ERROR"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/space_shuttle.toml b/ext/MetaData/references/space_shuttle.toml
index 15598138..69a95156 100644
--- a/ext/MetaData/references/space_shuttle.toml
+++ b/ext/MetaData/references/space_shuttle.toml
@@ -6,55 +6,55 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, w = 203000.0, g₀ = 32.174, ρ₀ = 0.002378, hᵣ = 23800, Rₑ = 20902900, μ = 1.4076539e16, S = 2690, a₀ = -0.20704, a₁ = 0.029244, b₀ = 0.07854, b₁ = -0.0061592, b₂ = 0.000621408, tf_l = 1500.0, tf_u = 2500.0, h_t0 = 2.6, ϕ_t0 = 0.0, θ_t0 = 0.0, v_t0 = 2.56, γ_t0 = -0.017453292519943295, ψ_t0 = 1.5707963267948966, α_t0 = 0.0, α_s = 0.0, β_s = 0.0, h_tf = 0.8, v_tf = 0.25, γ_tf = -0.08726646259971647, h_l = 0, ϕ_l = -6.283185307179586, ϕ_u = 6.283185307179586, θ_l = -1.5533430342749532, θ_u = 1.5533430342749532, v_l = 0.0001, γ_l = -1.5533430342749532, γ_u = 1.5533430342749532, ψ_l = -6.283185307179586, ψ_u = 6.283185307179586, α_l = -1.5707963267948966, α_u = 1.5707963267948966, β_l = -1.5533430342749532, β_u = 0.017453292519943295)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 6.232111392599022e-10
-value = -0.5958750134583887
+value = -0.5958750133508789
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
     modeling = "JuMP"
-    objective = -0.5958750133605147
+    objective = -0.5958750133627942
     platform = "cron-JuMP-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt"
     modeling = "JuMP"
-    objective = -0.5958750133605122
+    objective = -0.5958750133627957
     platform = "cron-JuMP-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = -0.5958750133605153
+    objective = -0.5958750133627942
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = -0.5958750133605122
+    objective = -0.5958750133627957
     platform = "cron-JuMP-robustness-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = -0.5958750133130643
+    objective = -0.5958750133150453
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = -0.5958750133130637
+    objective = -0.5958750133150452
     platform = "cron-JuMP-robustness-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = -0.5958750133508761
+    objective = -0.5958750133508767
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = -0.5958750133508751
+    objective = -0.595875013350878
     platform = "cron-JuMP-robustness-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = -0.5958750133627926
+    objective = -0.5958750133627936
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -99,7 +99,7 @@ value = -0.5958750134583887
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControl"
-    objective = -0.5958750133150457
+    objective = -0.595875013315046
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -144,12 +144,12 @@ value = -0.5958750134583887
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = -0.5958750133150457
+    objective = -0.595875013315046
     platform = "cron-OptimalControlScalarized-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = -0.5958750133150457
+    objective = -0.595875013315046
     platform = "cron-OptimalControlScalarized-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -159,5 +159,48 @@ value = -0.5958750134583887
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = -0.5958750134583887
+    objective = -0.5958750133508789
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "0f6b1ec2b60b46fcc5141592fd08794531e1f519"
+        configuration = "modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 106
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = -0.5958750133508789
+        os = "Linux"
+        platform = "cron-OptimalControl-robustness-1.13-ubuntu-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
+        solver = "ipopt"
+        status = "first_order"
+        time = 5.34838
+        violation = 3.1283448054253427e-11
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_init = 0.001
+            mu_strategy = "monotone"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/steering.toml b/ext/MetaData/references/steering.toml
index a1b063e4..fbc040e8 100644
--- a/ext/MetaData/references/steering.toml
+++ b/ext/MetaData/references/steering.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, a = 100, u_min = -1.5707963267948966, u_max = 1.5707963267948966, tf_l = 0, x₁_t0 = 0, x₂_t0 = 0, x₃_t0 = 0, x₄_t0 = 0, x₂_tf = 5, x₃_tf = 45, x₄_tf = 0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
-spread = 2.3955468295548094e-10
-value = 0.554571860956095
+spread = 2.395548831501e-10
+value = 0.5545718610889453
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -19,17 +19,17 @@ value = 0.554571860956095
     [[trapeze.results]]
     configuration = "solver=ipopt"
     modeling = "JuMP"
-    objective = 0.5545718610889453
+    objective = 0.5545718610889454
     platform = "cron-JuMP-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = 0.5545718610889453
+    objective = 0.5545718610889452
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = 0.5545718610889453
+    objective = 0.5545718610889454
     platform = "cron-JuMP-robustness-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -39,7 +39,7 @@ value = 0.554571860956095
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 0.5545718610887032
+    objective = 0.5545718610887033
     platform = "cron-JuMP-robustness-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
@@ -49,7 +49,7 @@ value = 0.554571860956095
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = 0.5545718609572946
+    objective = 0.5545718609572947
     platform = "cron-JuMP-robustness-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -159,5 +159,47 @@ value = 0.554571860956095
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = 0.554571860956095
+    objective = 0.5545718610889453
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "636fa809ae3e6c4367343be85d25d63fe119b5bd"
+        configuration = "modeler=adnlp solver=ipopt"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 8
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = 0.5545718610889453
+        os = "NT"
+        platform = "cron-OptimalControl-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573"
+        solver = "ipopt"
+        status = "first_order"
+        time = 0.227696
+        violation = 1.6216486486975157e-11
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_strategy = "adaptive"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/ext/MetaData/references/vanderpol.toml b/ext/MetaData/references/vanderpol.toml
index c0189114..3a8e4bf5 100644
--- a/ext/MetaData/references/vanderpol.toml
+++ b/ext/MetaData/references/vanderpol.toml
@@ -6,10 +6,10 @@
 agreement = true
 grid_size = 500
 parameters = "(t0 = 0, tf = 2, ω = 1, ε = 1, x₁_t0 = 1, x₂_t0 = 0)"
-run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288933404 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288939786 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288946781 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42288952545"
+run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387108959 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387112175 https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387115100"
 source = "computed"
 spread = 4.026341842688888e-15
-value = 1.0478115516287476
+value = 1.0478115516287507
 
     [[trapeze.results]]
     configuration = "solver=ipopt"
@@ -24,7 +24,7 @@ value = 1.0478115516287476
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
     modeling = "JuMP"
-    objective = 1.0478115516287516
+    objective = 1.0478115516287514
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_init=0.001"
@@ -34,7 +34,7 @@ value = 1.0478115516287476
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
     modeling = "JuMP"
-    objective = 1.0478115516287516
+    objective = 1.0478115516287514
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone"
@@ -44,7 +44,7 @@ value = 1.0478115516287476
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
     modeling = "JuMP"
-    objective = 1.0478115516287516
+    objective = 1.0478115516287514
     platform = "cron-JuMP-robustness-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "solver=ipopt mu_strategy=monotone mu_init=0.001"
@@ -54,7 +54,7 @@ value = 1.0478115516287476
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
     modeling = "OptimalControl"
-    objective = 1.0478115516287518
+    objective = 1.0478115516287512
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=adnlp solver=ipopt"
@@ -99,7 +99,7 @@ value = 1.0478115516287476
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControl"
-    objective = 1.0478115516287503
+    objective = 1.0478115516287476
     platform = "cron-OptimalControl-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -144,12 +144,12 @@ value = 1.0478115516287476
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = 1.0478115516287503
+    objective = 1.0478115516287476
     platform = "cron-OptimalControlScalarized-1.13-ubuntu-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
     modeling = "OptimalControlScalarized"
-    objective = 1.0478115516287503
+    objective = 1.0478115516287476
     platform = "cron-OptimalControlScalarized-1.13-windows-latest"
     [[trapeze.results]]
     configuration = "modeler=exa solver=madnlp"
@@ -159,5 +159,47 @@ value = 1.0478115516287476
 
     [[trapeze.solutions]]
     configurations = ["JuMP solver=ipopt", "JuMP solver=ipopt mu_init=0.001", "JuMP solver=ipopt mu_strategy=monotone", "JuMP solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt", "OptimalControl modeler=adnlp solver=ipopt mu_init=0.001", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone", "OptimalControl modeler=adnlp solver=ipopt mu_strategy=monotone mu_init=0.001", "OptimalControl modeler=exa solver=madnlp", "OptimalControl modeler=exa solver=madnlp barrier=loqo", "OptimalControl modeler=exa solver=madnlp mu_init=0.001", "OptimalControl modeler=exa solver=madnlp mu_init=1.0e-5", "OptimalControlScalarized modeler=exa solver=madnlp"]
-    objective = 1.0478115516287476
+    objective = 1.0478115516287507
     solves = 29
+
+        [trapeze.solutions.reference]
+        arch = "x86_64"
+        commit = "636fa809ae3e6c4367343be85d25d63fe119b5bd"
+        configuration = "modeler=adnlp solver=ipopt"
+        cpu = "AMD EPYC 7763 64-Core Processor"
+        iterations = 3
+        julia = "1.13.3"
+        modeler = "adnlp"
+        modeling = "OptimalControl"
+        objective = 1.0478115516287507
+        os = "NT"
+        platform = "cron-OptimalControl-1.13-windows-latest"
+        run = "https://github.com/control-toolbox/OptimalControlProblems.jl/actions/runs/42387105573"
+        solver = "ipopt"
+        status = "first_order"
+        time = 0.0753526
+        violation = 1.0363609970198695e-10
+
+            [trapeze.solutions.reference.options]
+            max_iter = 1000
+            max_wall_time = 500.0
+            mu_strategy = "adaptive"
+            print_level = 0
+            sb = "yes"
+            tol = 1.0e-8
+
+            [trapeze.solutions.reference.versions]
+            ADNLPModels = "0.8.15"
+            CTBase = "0.30.13"
+            CTDirect = "1.1.8-beta"
+            CTModels = "0.19.8"
+            CTSolvers = "0.5.8"
+            ExaModels = "0.12.2"
+            Ipopt_jll = "300.1400.1902+0"
+            MUMPS_seq_jll = "500.900.100+0"
+            MadNLP = "0.10.3"
+            NLPModels = "0.21.12"
+            NLPModelsIpopt = "0.11.3"
+            OptimalControl = "2.2.1-beta"
+            OptimalControlProblems = "0.5.0"
+            Test = "1.11.0"
diff --git a/src/README.md b/src/README.md
index 64e62572..a7755dac 100644
--- a/src/README.md
+++ b/src/README.md
@@ -13,7 +13,7 @@ In dependency order: a submodule uses only those above it (`MOD7`).
 | --- | --- |
 | `Modelings` | the backend types and the modeling table `MODELING_SPECS`, the single place where a modeling is declared |
 | `Metadata` | the list of the problems (the files of `ext/MetaData/`), the schema `METADATA_SCHEMA`, loading and checking the metadata, reading it (`problem_metadata`, `parameters_data`, `initial_guess_data`…), the analytic solutions used by the metadata files |
-| `References` | the reference objectives and known solutions (`ext/MetaData/references/`), `reference_objective`, `solution_kind`, `reference_solution_data`, `SAME_SOLUTION` |
+| `References` | the reference objectives and known solutions (`ext/MetaData/references/`), `reference_objective`, `solution_kind`, `reference_solution_data`, `SAME_SOLUTION`; the references as values (#409): `AbstractReference`, `NumericalReference` (a known solution and how it was obtained), `AnalyticReference`, `references`, `best_reference`, `analytic_reference`, `objective(ref)` |
 | `Catalogue` | one function per problem (`beam`…), whose methods the extensions define, `problems`, `status`, the fingerprint of the files that define a problem |
 | `Problems` | `Problem`, `features`, `solution`, and the accessors shared with OptimalControl (`state`, `objective`…) |
 
diff --git a/src/References/References.jl b/src/References/References.jl
index a72e4d17..ee5b721d 100644
--- a/src/References/References.jl
+++ b/src/References/References.jl
@@ -2,19 +2,27 @@
 The references of the problems: the reference objectives and the known solutions of each
 problem, read from `ext/MetaData/references/<problem>.toml` (generated by
 `tools/references/update.jl`), and the analytic solutions of the continuous problems:
-`reference_objective`, `solution_kind`, `reference_solution_data`, `SAME_SOLUTION`.
+`reference_objective`, `solution_kind`, `reference_solution_data`, `SAME_SOLUTION`. Each
+known solution says how it was obtained, as a `NumericalReference`; an analytic solution is
+an `AnalyticReference` (#409): `references`, `best_reference`, `analytic_reference`, and
+`objective(ref)`.
 
 Depends on `Metadata`, whose metadata it completes with the field `:reference_objective`.
 """
 module References
 
-using DocStringExtensions: TYPEDSIGNATURES
+using CTBase: CTBase
+using CTModels: CTModels
+using DocStringExtensions: TYPEDEF, TYPEDFIELDS, TYPEDSIGNATURES
 using TOML: TOML
 
 using ..Metadata: Metadata
 
+include(joinpath(@__DIR__, "types.jl"))
 include(joinpath(@__DIR__, "objectives.jl"))
 
 export reference_objective, reference_solution_data, solution_kind, SAME_SOLUTION
+export AbstractReference, AnalyticReference, NumericalReference
+export references, best_reference, analytic_reference
 
 end # module References
diff --git a/src/References/objectives.jl b/src/References/objectives.jl
index cc320a62..c39ef90c 100644
--- a/src/References/objectives.jl
+++ b/src/References/objectives.jl
@@ -45,7 +45,20 @@ function _references(
             )
             return REFERENCE_FIELDS[key](table[string(key)])
         end
-        push!(entries, Symbol(scheme) => NamedTuple{keys(REFERENCE_FIELDS)}(values))
+        # the known solutions with how they were obtained (#409), best first; none for a
+        # file written before #409
+        found = [
+            _numerical_reference(problem, Symbol(scheme), Int(table["grid_size"]), s) for
+            s in table["solutions"]
+        ]
+        references = NumericalReference[r for r in found if !isnothing(r)]
+        length(references) == length(found) || empty!(references)
+        push!(
+            entries,
+            Symbol(scheme) => NamedTuple{(keys(REFERENCE_FIELDS)..., :references)}((
+                values..., references
+            )),
+        )
     end
     return NamedTuple(entries)
 end
@@ -133,3 +146,63 @@ function solution_kind(
     any(s -> _relative_gap(objective, s) <= SAME_SOLUTION, solutions) && return :local
     return :new
 end
+
+"""
+$(TYPEDSIGNATURES)
+
+The known solutions of `problem` discretized on `grid_size` steps with `scheme`, best first,
+each with how it was obtained (#409); empty when the problem has no reference for this
+discretization. The first one is the reference objective.
+
+# Example
+
+```julia-repl
+julia> ref = first(OptimalControlProblems.references(:beam))
+julia> ref.solver, ref.options.tol, ref.status
+```
+"""
+function references(
+    problem::Symbol;
+    grid_size::Int=Metadata.grid_size_data(problem),
+    scheme::Symbol=:trapeze,
+)
+    tables = get(Metadata.problem_metadata(problem), :reference_objective, nothing)
+    isnothing(tables) && return NumericalReference[]
+    entry = get(tables, scheme, nothing)
+    (isnothing(entry) || entry.grid_size != grid_size) && return NumericalReference[]
+    return entry.references
+end
+
+"""
+$(TYPEDSIGNATURES)
+
+The best known solution of `problem` discretized on `grid_size` steps with `scheme`, with
+how it was obtained, or `nothing`: its objective is `reference_objective(problem)`.
+"""
+function best_reference(
+    problem::Symbol;
+    grid_size::Int=Metadata.grid_size_data(problem),
+    scheme::Symbol=:trapeze,
+)
+    found = references(problem; grid_size, scheme)
+    return isempty(found) ? nothing : first(found)
+end
+
+"""
+$(TYPEDSIGNATURES)
+
+The analytic solution of the continuous `problem` for `parameters` (merged with the default
+ones) as an [`AnalyticReference`](@ref), or `nothing` when none is known for them.
+"""
+function analytic_reference(problem::Symbol, parameters::Union{Nothing,NamedTuple}=nothing)
+    solution = reference_solution_data(problem, parameters)
+    isnothing(solution) && return nothing
+    return AnalyticReference(
+        problem,
+        Metadata.parameters_data(problem, parameters),
+        Float64(solution.objective),
+        solution.state,
+        solution.control,
+        solution.source,
+    )
+end
diff --git a/src/References/types.jl b/src/References/types.jl
new file mode 100644
index 00000000..9b5a49ed
--- /dev/null
+++ b/src/References/types.jl
@@ -0,0 +1,180 @@
+# The references of a problem as values (#409): an analytic solution of the continuous
+# problem, or a solve of the discretized problem with how it was obtained. Built from the
+# metadata and from the generated reference files, never written by hand.
+
+"""
+$(TYPEDEF)
+
+A reference solution of a problem: its objective, read with `objective(ref)`.
+
+Two kinds, with different data: [`AnalyticReference`](@ref), the solution of the continuous
+problem, and [`NumericalReference`](@ref), a solve of the discretized problem.
+"""
+abstract type AbstractReference end
+
+"""
+$(TYPEDSIGNATURES)
+
+The objective of the reference `ref`. Every subtype of `AbstractReference` implements it.
+"""
+function CTModels.Components.objective(ref::AbstractReference)
+    return throw(
+        CTBase.Exceptions.NotImplemented(
+            "objective is not implemented for $(typeof(ref))";
+            required_method="objective(::$(typeof(ref)))",
+            suggestion="define it for every subtype of AbstractReference",
+        ),
+    )
+end
+
+"""
+$(TYPEDEF)
+
+The analytic solution of the continuous problem `problem` for `parameters` (#379): `state`
+and `control`, functions of the time, the `objective`, and the `source` it comes from. The
+objective of a discretized problem differs from it by the discretization error.
+
+# Fields
+
+$(TYPEDFIELDS)
+"""
+struct AnalyticReference{S,U} <: AbstractReference
+    "Name of the problem."
+    problem::Symbol
+    "Parameters of the problem the solution is for."
+    parameters::Union{Nothing,NamedTuple}
+    "Objective of the continuous problem."
+    objective::Float64
+    "State, a function of the time."
+    state::S
+    "Control, a function of the time."
+    control::U
+    "Where the solution comes from (a publication)."
+    source::String
+end
+
+CTModels.Components.objective(ref::AnalyticReference) = ref.objective
+
+"""
+$(TYPEDEF)
+
+A known solution of the problem `problem` discretized on `grid_size` steps with `scheme`
+(#378), and how it was obtained (#409): one of the solves that reached it, chosen to
+represent it (a strict convergence before an acceptable one, then the smallest constraint
+violation), with its modeling, modeler, solver and every option of the solver, its exit
+status, iterations, time and constraint violation, and the environment of its run.
+
+Generated by `tools/references/update.jl` from the convergence and robustness tests. The
+trajectories are not kept: the configuration is enough to solve again.
+
+# Fields
+
+$(TYPEDFIELDS)
+"""
+struct NumericalReference <: AbstractReference
+    "Name of the problem."
+    problem::Symbol
+    "Discretization scheme, e.g. `:trapeze`."
+    scheme::Symbol
+    "Number of steps of the grid."
+    grid_size::Int
+    "Objective of the solve."
+    objective::Float64
+    "Number of solves, every modeling, configuration and platform, that reached the solution."
+    solves::Int
+    "The configurations that reached it, `\"<modeling> <configuration>\"`."
+    configurations::Vector{String}
+    "Modeling of the solve, e.g. `:OptimalControl`."
+    modeling::Symbol
+    "Configuration of the solve, as the tests name it."
+    configuration::String
+    "Modeler of the solve (`:adnlp`, `:exa`), `nothing` for JuMP."
+    modeler::Union{Nothing,Symbol}
+    "Solver of the solve, e.g. `:ipopt`."
+    solver::Symbol
+    "Every option given to the solver, defaults included."
+    options::NamedTuple
+    "Exit status of the solver, e.g. `:first_order`, `:LOCALLY_SOLVED`, `:acceptable`."
+    status::Symbol
+    "Number of iterations."
+    iterations::Int
+    "Solve time in seconds, on `platform`."
+    time::Float64
+    "Largest violation of the constraints, as the modeling measures it."
+    violation::Float64
+    "Platform of the run, e.g. `cron-OptimalControl-1.13-ubuntu-latest`."
+    platform::String
+    "The run that solved it (a GitHub Actions run), or `local`."
+    run::String
+    "Commit of the package."
+    commit::String
+    "Julia version."
+    julia::VersionNumber
+    "Operating system, architecture and processor of the platform."
+    machine::NamedTuple{(:os, :arch, :cpu),Tuple{String,String,String}}
+    "Versions of the packages that modeled, discretized and solved it."
+    versions::NamedTuple
+end
+
+CTModels.Components.objective(ref::NumericalReference) = ref.objective
+
+# A TOML table as a `NamedTuple` with sorted keys, so that two equal tables give equal values.
+_namedtuple(table) = (; (Symbol(k) => table[k] for k in sort!(collect(keys(table))))...)
+
+# The `NumericalReference` of a known solution, from its table in a reference file, or
+# `nothing` for a file written before #409. Such a file still loads, so that
+# `tools/references/update.jl`, which loads the package, can write it again; the consistency
+# tests fail on it.
+function _numerical_reference(problem::Symbol, scheme::Symbol, grid_size::Int, solution)
+    haskey(solution, "reference") || return nothing
+    r = solution["reference"]
+    return NumericalReference(
+        problem,
+        scheme,
+        grid_size,
+        Float64(solution["objective"]),
+        Int(solution["solves"]),
+        String.(solution["configurations"]),
+        Symbol(r["modeling"]),
+        r["configuration"],
+        isempty(r["modeler"]) ? nothing : Symbol(r["modeler"]),
+        Symbol(r["solver"]),
+        _namedtuple(r["options"]),
+        Symbol(r["status"]),
+        Int(r["iterations"]),
+        Float64(r["time"]),
+        Float64(r["violation"]),
+        r["platform"],
+        r["run"],
+        r["commit"],
+        VersionNumber(r["julia"]),
+        (os=r["os"], arch=r["arch"], cpu=r["cpu"]),
+        _namedtuple(r["versions"]),
+    )
+end
+
+function Base.show(io::IO, ref::NumericalReference)
+    return print(io, "NumericalReference(:", ref.problem, ", ", ref.objective, ")")
+end
+
+function Base.show(io::IO, ::MIME"text/plain", ref::NumericalReference)
+    println(
+        io,
+        "Known solution of ",
+        ref.problem,
+        ", ",
+        ref.scheme,
+        " on ",
+        ref.grid_size,
+        " steps",
+    )
+    println(io, "  objective:  ", ref.objective, " (", ref.solves, " solves)")
+    modeler = isnothing(ref.modeler) ? "" : string(ref.modeler, " + ")
+    println(io, "  solved by:  ", ref.modeling, ", ", modeler, ref.solver, ", ", ref.status)
+    println(io, "  iterations: ", ref.iterations, ", violation ", ref.violation)
+    return print(io, "  run:        ", ref.run, " (", ref.platform, ")")
+end
+
+function Base.show(io::IO, ref::AnalyticReference)
+    return print(io, "AnalyticReference(:", ref.problem, ", ", ref.objective, ")")
+end
diff --git a/test/suite/core/test_consistency.jl b/test/suite/core/test_consistency.jl
index a545b1af..d652851b 100644
--- a/test/suite/core/test_consistency.jl
+++ b/test/suite/core/test_consistency.jl
@@ -87,6 +87,10 @@ is reported, not failed (the comparison rule fails when it has no other referenc
 """
 function test_reference_objectives()
     Test.@testset "reference objectives" verbose = VERBOSE begin
+        # generated by tools/references/update.jl, never written by hand: its header (#409)
+        for file in readdir(OCP.References.REFERENCES_PATH; join=true)
+            Test.@test startswith(read(file, String), "# Reference objectives of the problem")
+        end
         for problem in OCP.problems(; status=:all)
             data = OCP.problem_metadata(problem)
             references = get(data, :reference_objective, (;))
@@ -105,6 +109,11 @@ function test_reference_objectives()
                 Test.@test all(s -> kind(s) === :local, entry.solutions[2:end])
                 far = maximum(abs, entry.solutions) * 10 + 1
                 Test.@test kind(far) === :new
+                # each known solution says how it was obtained (#409): a file written before
+                # has none, compute the references again (tools/references/update.jl)
+                Test.@test length(entry.references) == length(entry.solutions)
+                Test.@test [r.objective for r in entry.references] == entry.solutions
+                Test.@test all(r -> r.grid_size == entry.grid_size && r.scheme === scheme, entry.references)
             end
             valid = [name for (name, s) in pairs(OCP.status(problem)) if s === :valid]
             (isempty(valid) || !isnothing(OCP.reference_objective(problem))) && continue
diff --git a/test/suite/core/test_references.jl b/test/suite/core/test_references.jl
index bc02ef9a..e709296a 100644
--- a/test/suite/core/test_references.jl
+++ b/test/suite/core/test_references.jl
@@ -1,5 +1,6 @@
 module TestReferences
 
+using CTBase: CTBase
 using Test: Test
 using OptimalControlProblems: OptimalControlProblems
 
@@ -7,34 +8,114 @@ const VERBOSE = isdefined(Main, :TestOptions) ? Main.TestOptions.VERBOSE : true
 const SHOWTIMING = isdefined(Main, :TestOptions) ? Main.TestOptions.SHOWTIMING : true
 
 const OCP = OptimalControlProblems
-const TOOL = joinpath(pkgdir(OptimalControlProblems), "tools", "references", "references.jl")
+const TOOL = joinpath(
+    pkgdir(OptimalControlProblems), "tools", "references", "references.jl"
+)
 isdefined(@__MODULE__, :ReferenceObjectives) || include(TOOL)
 
+# A reference type that does not implement the contract (TST3: at the top level).
+struct FakeReference <: OptimalControlProblems.AbstractReference end
+
 # Fake problems: the tool is tested on its own, apart from the collection.
 const DEFAULTS = Dict(
     :alpha => (grid_size=100, parameters="(a = 1,)"),
     :omega => (grid_size=50, parameters="nothing"),
 )
-const HEADER = "test\tproblem\tmodeling\tconfiguration\tgrid_size\tscheme\tresult\treason\tobjective\titerations\ttime\tgap"
+const HEADER = "test\tproblem\tmodeling\tconfiguration\tgrid_size\tscheme\tresult\treason\tobjective\titerations\ttime\tgap\tmodeler\tsolver\toptions\tviolation"
 
-row(test, problem, modeling, grid, result, objective) = join(
-    (test, problem, modeling, "c=1", grid, "trapeze", result, "R", objective, "3", "0.1", ""), "\t"
+# A row of a results file; `reason` and `violation` choose the representative of a solution
+# (#409): a strict convergence before an acceptable one, then the smallest violation.
+function row(
+    test, problem, modeling, grid, result, objective; reason="R", violation="1.0e-9"
 )
+    return join(
+        (
+            test,
+            problem,
+            modeling,
+            "c=1",
+            grid,
+            "trapeze",
+            result,
+            reason,
+            objective,
+            "3",
+            "0.1",
+            "",
+            "adnlp",
+            "ipopt",
+            "{tol = 1.0e-8, mu_strategy = \"adaptive\"}",
+            violation,
+        ),
+        "\t",
+    )
+end
+
+# The environment of a run, written next to its results file by the tests (#412).
+function environment(file; run="https://run/1")
+    return write(
+        replace(file, "-results.tsv" => "-environment.toml"),
+        """
+        arch = "x86_64"
+        commit = "abc"
+        cpu = "CPU"
+        julia = "1.13.3"
+        os = "Linux"
+        run = "$(run)"
+        threads = 1
+
+        [packages]
+        Ipopt_jll = "300.1400.1902+0"
+        """,
+    )
+end
 
 function results_dir()
     dir = mktempdir()
     mkpath(joinpath(dir, "cron-ubuntu"))
+    file = joinpath(dir, "cron-ubuntu", "X-cron-results.tsv")
+    environment(file)
     write(
-        joinpath(dir, "cron-ubuntu", "X-cron-results.tsv"),
+        file,
         join(
             [
                 HEADER,
-                row("test_convergence c=1", "alpha", "A", "100", "converged", "2.0000000001"),
-                row("test_convergence c=1", "alpha", "B", "100", "converged", "2.0"),
+                # the smallest objective and violation, but acceptable: not the representative
+                row(
+                    "test_convergence c=1",
+                    "alpha",
+                    "A",
+                    "100",
+                    "converged",
+                    "1.99999999";
+                    reason="acceptable",
+                    violation="1.0e-13",
+                ),
+                row(
+                    "test_convergence c=1",
+                    "alpha",
+                    "B",
+                    "100",
+                    "converged",
+                    "2.0";
+                    violation="1.0e-6",
+                ),
+                # strict, the smallest violation: the representative of the solution
+                row(
+                    "test_convergence c=1",
+                    "alpha",
+                    "G",
+                    "100",
+                    "converged",
+                    "2.0000000001";
+                    violation="1.0e-12",
+                ),
                 row("test_convergence c=1", "alpha", "C", "100", "not_converged", "1.0"), # not converged
                 row("test_convergence c=1", "alpha", "D", "10", "converged", "1.5"), # not the default grid
                 row("test_smoke c=1", "alpha", "E", "100", "converged", "1.0"), # not a convergence test
-                row("test_robustness c=1 mu=1", "alpha", "F", "100", "converged", "2.00001"), # same solution
+                row(
+                    "test_robustness c=1 mu=1", "alpha", "F", "100", "converged", "2.00001"
+                ), # same solution
                 row("test_convergence c=1", "omega", "A", "50", "converged", "3.0"),
                 row("test_convergence c=1", "omega", "B", "50", "converged", "3.3"),
                 row("test_convergence c=1", "gamma", "A", "50", "converged", "1.0"), # unknown problem
@@ -42,7 +123,14 @@ function results_dir()
             "\n",
         ) * "\n",
     )
-    write(joinpath(dir, "local-results.tsv"), HEADER * "\n" * row("test_convergence c=1", "omega", "C", "50", "converged", "3.0") * "\n")
+    write(
+        joinpath(dir, "local-results.tsv"),
+        HEADER *
+        "\n" *
+        row("test_convergence c=1", "omega", "C", "50", "converged", "3.0") *
+        "\n",
+    )
+    environment(joinpath(dir, "local-results.tsv"); run="local")
     return dir
 end
 
@@ -50,19 +138,36 @@ function test_references()
     Test.@testset "reference objectives" verbose = VERBOSE showtiming = SHOWTIMING begin
         R = ReferenceObjectives
         rows = R.read_results([results_dir()])
-        Test.@test length(rows) == 10
-        Test.@test count(r -> r["platform"] == "cron-ubuntu", rows) == 9
+        Test.@test length(rows) == 11
+        Test.@test count(r -> r["platform"] == "cron-ubuntu", rows) == 10
         Test.@test count(r -> r["platform"] == "local", rows) == 1
 
         tables = R.references(rows, DEFAULTS; run="https://run")
         Test.@test sort!(collect(keys(tables))) == [:alpha, :omega]
         alpha = tables[:alpha]["trapeze"]
-        # the best converged objective at the default grid, from convergence and robustness tests
-        Test.@test alpha["value"] == 2.0
-        Test.@test length(alpha["results"]) == 3
+        # the objective of the representative of the best solution, at the default grid, from
+        # convergence and robustness tests (#409): strict, then the smallest violation
+        Test.@test alpha["value"] == 2.0000000001
+        Test.@test length(alpha["results"]) == 4
         # within SAME_SOLUTION (1e-5): one solution, the configurations agree
         Test.@test length(alpha["solutions"]) == 1
-        Test.@test alpha["solutions"][1]["solves"] == 3
+        Test.@test alpha["solutions"][1]["solves"] == 4
+        reference = alpha["solutions"][1]["reference"]
+        Test.@test reference["modeling"] == "G"
+        Test.@test reference["objective"] == alpha["value"]
+        Test.@test reference["violation"] == 1e-12
+        Test.@test reference["status"] == "R"
+        Test.@test reference["options"] == Dict("tol" => 1e-8, "mu_strategy" => "adaptive")
+        Test.@test (reference["modeler"], reference["solver"]) == ("adnlp", "ipopt")
+        Test.@test reference["run"] == "https://run/1" && reference["commit"] == "abc"
+        Test.@test reference["platform"] == "cron-ubuntu"
+        Test.@test reference["versions"] == Dict("Ipopt_jll" => "300.1400.1902+0")
+        # without a strict convergence, an acceptable solve represents its solution
+        acceptable = R.representative([
+            merge(r, Dict("reason" => "acceptable")) for
+            r in rows if r["problem"] == "omega"
+        ])
+        Test.@test acceptable["reason"] == "acceptable"
         Test.@test alpha["agreement"]
         Test.@test alpha["grid_size"] == 100 && alpha["parameters"] == "(a = 1,)"
         omega = tables[:omega]["trapeze"]
@@ -80,23 +185,88 @@ function test_references()
         Test.@test basename.(files) == ["alpha.toml", "omega.toml"]
         Test.@test startswith(read(files[1], String), "# Reference objectives")
         read_back = OCP.References._references(:alpha, files[1])
-        Test.@test read_back.trapeze.value == 2.0
+        Test.@test read_back.trapeze.value == 2.0000000001
         Test.@test read_back.trapeze.source === :computed
         Test.@test read_back.trapeze.run == "https://run"
-        Test.@test read_back.trapeze.solutions == [2.0]
-        Test.@test OCP.References._references(:omega, files[2]).trapeze.solutions == [3.0, 3.3]
+        Test.@test read_back.trapeze.solutions == [2.0000000001]
+        # the known solution with how it was obtained (#409)
+        ref = only(read_back.trapeze.references)
+        Test.@test ref isa OCP.NumericalReference
+        Test.@test (ref.problem, ref.scheme, ref.grid_size) == (:alpha, :trapeze, 100)
+        Test.@test OCP.objective(ref) == 2.0000000001
+        Test.@test (ref.modeling, ref.modeler, ref.solver, ref.status) ==
+            (:G, :adnlp, :ipopt, :R)
+        Test.@test ref.options == (mu_strategy="adaptive", tol=1e-8)
+        Test.@test ref.julia == v"1.13.3" && ref.machine.os == "Linux"
+        Test.@test ref.versions.Ipopt_jll == "300.1400.1902+0"
+        Test.@test ref.solves == 4 && ref.violation == 1e-12
+        Test.@test OCP.References._references(:omega, files[2]).trapeze.solutions ==
+            [3.0, 3.3]
         # a scheme not recomputed is kept
         R.write_references(dir, Dict(:alpha => Dict{String,Any}("midpoint" => alpha)))
-        Test.@test keys(OCP.References._references(:alpha, files[1])) ⊇ (:trapeze, :midpoint)
+        Test.@test keys(OCP.References._references(:alpha, files[1])) ⊇
+            (:trapeze, :midpoint)
 
         # a file that misses a field is an error at load time
         broken = joinpath(dir, "broken.toml")
         write(broken, "[trapeze]\nvalue = 1.0\n")
         Test.@test_throws ErrorException OCP.References._references(:alpha, broken)
         Test.@test isnothing(OCP.References._references(:alpha, joinpath(dir, "none.toml")))
+
+        # a file written before #409 loads, without references, so that the tool can write
+        # it again; the consistency tests fail on it
+        old = joinpath(dir, "old.toml")
+        table = deepcopy(alpha)
+        foreach(s -> delete!(s, "reference"), table["solutions"])
+        R.write_references(dir, Dict(:old => Dict{String,Any}("trapeze" => table)))
+        Test.@test isempty(OCP.References._references(:old, old).trapeze.references)
+
+        # results recorded before #409, or without their environment, cannot say how a
+        # solution was obtained: an error
+        stale = mktempdir()
+        write(
+            joinpath(stale, "X-results.tsv"), join(split(HEADER, '\t')[1:12], '\t') * "\n"
+        )
+        Test.@test_throws ErrorException R.read_results([stale])
+        bare = mktempdir()
+        write(joinpath(bare, "X-results.tsv"), HEADER * "\n")
+        Test.@test_throws ErrorException R.read_results([bare])
+    end
+end
+
+# The references of the collection as values (#409): known solutions and analytic ones.
+function test_reference_api()
+    Test.@testset "reference values" verbose = VERBOSE showtiming = SHOWTIMING begin
+        Test.@test_throws CTBase.Exceptions.NotImplemented OCP.objective(FakeReference())
+        for problem in OCP.problems(; status=:all)
+            found = OCP.references(problem)
+            best = OCP.best_reference(problem)
+            if isempty(found)
+                Test.@test isnothing(best)
+                Test.@test isnothing(OCP.reference_objective(problem))
+            else
+                Test.@test best === first(found)
+                Test.@test OCP.objective(best) == OCP.reference_objective(problem)
+                Test.@test issorted(OCP.objective.(found))
+                Test.@test all(r -> r.problem === problem, found)
+                Test.@test all(
+                    r -> OCP.solution_kind(problem, OCP.objective(r)) in (:best, :local),
+                    found,
+                )
+            end
+            # another discretization has no reference
+            Test.@test isempty(OCP.references(problem; grid_size=1))
+            analytic = OCP.analytic_reference(problem)
+            data = OCP.reference_solution_data(problem)
+            Test.@test isnothing(analytic) == isnothing(data)
+            isnothing(analytic) && continue
+            Test.@test analytic isa OCP.AnalyticReference
+            Test.@test OCP.objective(analytic) == data.objective
+            Test.@test analytic.source == data.source
+        end
     end
 end
 
 end # module TestReferences
 
-test_references() = TestReferences.test_references()
+test_references() = (TestReferences.test_references(); TestReferences.test_reference_api())
diff --git a/tools/references/README.md b/tools/references/README.md
index f48fe526..35be1a9e 100644
--- a/tools/references/README.md
+++ b/tools/references/README.md
@@ -11,13 +11,41 @@ discretization, or `nothing`. `OptimalControlProblems.solution_kind(p, objective
 scheme)` says which known solution an objective is: `:best`, `:local` or `:new`. The
 convergence and robustness reports (`test/README.md`) show both for each solve.
 
+Each known solution also says **how it was obtained** (#409), as a value:
+`OptimalControlProblems.references(p; grid_size, scheme)` returns them best first, as
+`NumericalReference`s, and `OptimalControlProblems.best_reference(p)` the first one:
+
+```julia
+ref = OptimalControlProblems.best_reference(:moonlander)
+ref.modeling, ref.modeler, ref.solver   # how it was modeled and solved
+ref.options                             # every option of the solver, defaults included
+ref.status, ref.iterations, ref.violation
+ref.run, ref.commit, ref.versions       # where, and with which packages
+OptimalControlProblems.objective(ref)
+```
+
+An analytic solution of the continuous problem is an `AnalyticReference`
+(`OptimalControlProblems.analytic_reference(p, parameters)`), with its state, control,
+objective and source. Both are `AbstractReference`s. The trajectories of a numerical
+solution are not kept: the configuration is enough to solve again, and the trajectories and
+plots of the last runs belong to the results pipeline (#314).
+
 ## What a reference is
 
-- **Value:** the best objective among the solves that converged, over every modeling,
-  configuration and platform of the convergence and robustness tests, at the default grid.
-  Every problem of the collection is a minimization.
+- **Value:** the objective of the best solution among the solves that converged, over every
+  modeling, configuration and platform of the convergence and robustness tests, at the
+  default grid. Every problem of the collection is a minimization.
 - **`solutions`:** the distinct solutions reached, best first, each with its objective, its
-  number of solves and the configurations that reached it. Two objectives within a relative
+  number of solves, the configurations that reached it, and its **`reference`**: the solve
+  chosen to represent it, whose objective is the solution's. The representative is a strict
+  convergence before an acceptable one (`acceptable`, `ALMOST_LOCALLY_SOLVED`,
+  `SOLVED_TO_ACCEPTABLE_LEVEL`), then the solve with the smallest constraint violation, then
+  the smallest objective. Its table holds the modeling, configuration, modeler, solver and
+  every option, the exit status, iterations, time and violation, the platform, and the
+  environment of its run (run, commit, Julia, machine, package versions), read from
+  `<report>-environment.toml`. The violation is what each modeling measures (JuMP counts the
+  variable bounds, relaxed by Ipopt): comparing it across modelings favours the
+  OptimalControl ones when both are strict. Two objectives within a relative
   `OptimalControlProblems.SAME_SOLUTION` (1e-5) are the same solution. 1e-5 is above the
   accuracy of the solvers at `tol = 1e-8` (about 1e-6), and below the gap between the two
   known solutions of `mountain_car` (1.6e-4).
@@ -26,11 +54,16 @@ convergence and robustness reports (`test/README.md`) show both for each solve.
 - **`grid_size`, `parameters`:** the default grid and parameters of the problem when the
   reference was computed. If either changes, the consistency tests fail. Compute the
   reference again.
-- **`run`:** where the results come from (a cron run, or `local`). The package versions are
-  in the logs of that run.
+- **`run`:** where the results come from (the cron runs, or `local`). Each solution says which
+  run solved its representative, and the versions of its packages.
 
 ## Computing the references
 
+The results must say how each solve was configured (columns `modeler`, `solver`, `options`,
+`violation` and the file `<report>-environment.toml`, written by the tests since #412):
+older results are refused. A reference file written before #409 still loads, without
+`references`, so that this script can write it again; the consistency tests fail on it.
+
 From the cron runs: the weekly convergence runs (one per environment: JuMP, OptimalControl,
 OptimalControlScalarized) **and** the latest monthly robustness run (`CronRobustness.yml`),
 on `main` or on a branch with `workflow_dispatch`. Give all of them together: each file is
diff --git a/tools/references/references.jl b/tools/references/references.jl
index 1f092307..6fe19454 100644
--- a/tools/references/references.jl
+++ b/tools/references/references.jl
@@ -8,6 +8,11 @@
 # within a relative `OptimalControlProblems.SAME_SOLUTION` are the same solution. The
 # reference objective is the best (smallest: every problem of the collection is a
 # minimization); the configurations agree when they all reached one solution.
+#
+# Each solution says how it was obtained (#409): one of the solves that reached it, its
+# representative, with its modeling, modeler, solver and every option, its status, iterations,
+# time and constraint violation, and the environment of its run (`<report>-environment.toml`:
+# Julia, machine, commit, run, versions of the packages). Nothing is written by hand.
 module ReferenceObjectives
 
 using OptimalControlProblems: OptimalControlProblems
@@ -18,11 +23,17 @@ const SAME_SOLUTION = OptimalControlProblems.SAME_SOLUTION
 """The tests whose results count: convergence (#360) and robustness (#378)."""
 const TESTS = ("test_convergence", "test_robustness")
 
+# The columns that say how a solve was configured (#409), written by the tests since #412.
+const SETTINGS_COLUMNS = ("modeler", "solver", "options", "violation")
+
 """
     read_results(dirs) -> Vector{Dict{String,String}}
 
 Rows of every `*-results.tsv` under `dirs`, read by column name, with the `platform`: the
-first directory under the given one (the name of a CI artifact), or `local`.
+first directory under the given one (the name of a CI artifact), or `local`; and
+`environment`, the file `<report>-environment.toml` written by the same run, read with
+`environment`. A results file without the columns of #409 is an error: its solves cannot say
+how they were obtained.
 """
 function read_results(dirs)
     rows = Dict{String,String}[]
@@ -33,17 +44,85 @@ function read_results(dirs)
         lines = readlines(joinpath(root, file))
         isempty(lines) && continue
         header = split(lines[1], '\t')
+        missing_columns = setdiff(SETTINGS_COLUMNS, header)
+        isempty(missing_columns) || error(
+            "$(joinpath(root, file)) has no column $(join(missing_columns, ", ")): " *
+            "results recorded before #409, solve again",
+        )
+        environment = joinpath(root, replace(file, "-results.tsv" => "-environment.toml"))
+        isfile(environment) || error("$(environment) is missing: its results cannot say how they were obtained")
         for line in lines[2:end]
             cells = split(line, '\t')
             length(cells) == length(header) || continue
             row = Dict{String,String}(String(h) => String(c) for (h, c) in zip(header, cells))
             row["platform"] = platform
+            row["environment"] = environment
             push!(rows, row)
         end
     end
     return rows
 end
 
+"""The environment of the run of `row` (`<report>-environment.toml`), read once per file."""
+environment(row) = get!(() -> TOML.parsefile(row["environment"]), ENVIRONMENTS, row["environment"])
+const ENVIRONMENTS = Dict{String,Dict{String,Any}}()
+
+# Statuses of a solve that stopped at an acceptable point, not at the requested tolerance:
+# still a solution (#409), ranked after a strict convergence when one represents a solution.
+const ACCEPTABLE = ("acceptable", "ALMOST_LOCALLY_SOLVED", "SOLVED_TO_ACCEPTABLE_LEVEL")
+
+parse_violation(cell) = isempty(cell) ? Inf : (v = parse(Float64, cell); isnan(v) ? Inf : v)
+
+"""
+    representative(rows) -> Dict{String,String}
+
+The solve that represents a solution reached by `rows`: a strict convergence before an
+acceptable one, then the smallest constraint violation, then the smallest objective; ties
+are broken by modeling, configuration and platform, so that the choice does not depend on
+the order of the files.
+"""
+function representative(rows)
+    key(r) = (
+        r["reason"] in ACCEPTABLE,
+        parse_violation(r["violation"]),
+        parse(Float64, r["objective"]),
+        r["modeling"],
+        r["configuration"],
+        r["platform"],
+    )
+    return first(sort(rows; by=key))
+end
+
+"""
+    reference(row) -> Dict{String,Any}
+
+How the solve `row` was obtained, as the table `reference` of a solution: everything the
+results and the environment of its run say, so that it can be solved again.
+"""
+function reference(row)
+    env = environment(row)
+    return Dict{String,Any}(
+        "objective" => parse(Float64, row["objective"]),
+        "modeling" => row["modeling"],
+        "configuration" => row["configuration"],
+        "modeler" => row["modeler"],
+        "solver" => row["solver"],
+        "options" => TOML.parse("o = " * row["options"])["o"],
+        "status" => row["reason"],
+        "iterations" => parse(Int, row["iterations"]),
+        "time" => parse(Float64, row["time"]),
+        "violation" => parse_violation(row["violation"]),
+        "platform" => row["platform"],
+        "run" => env["run"],
+        "commit" => env["commit"],
+        "julia" => env["julia"],
+        "os" => env["os"],
+        "arch" => env["arch"],
+        "cpu" => env["cpu"],
+        "versions" => env["packages"],
+    )
+end
+
 """Relative spread of `values` around the smallest one (absolute when it is zero)."""
 function spread(values)
     best = minimum(values)
@@ -82,7 +161,9 @@ function references(rows, defaults; run::AbstractString)
             ) for r in sort(group; by=r -> (r["modeling"], r["configuration"], r["platform"]))
         ]
         get!(tables, problem, Dict{String,Any}())[scheme] = Dict{String,Any}(
-            "value" => minimum(objectives),
+            # the objective of the representative of the best solution: a solve that can be
+            # redone (#409), within `SAME_SOLUTION` of the smallest objective
+            "value" => first(found)["objective"],
             "grid_size" => defaults[problem].grid_size,
             "parameters" => defaults[problem].parameters,
             "source" => "computed",
@@ -101,8 +182,8 @@ end
 
 The distinct solutions reached by the converged `rows`, best first: consecutive objectives
 (sorted) within a relative `SAME_SOLUTION` of the first one of a group are the same solution.
-Each with its objective (the best of its group), its number of solves and the
-configurations that reached it.
+Each with the objective of its `representative`, its number of solves, the configurations
+that reached it, and how its representative was obtained (`reference`, #409).
 """
 function solutions(rows)
     objective(r) = parse(Float64, r["objective"])
@@ -116,13 +197,15 @@ function solutions(rows)
             push!(last(groups), r)
         end
     end
-    return [
-        Dict{String,Any}(
-            "objective" => objective(first(g)),
+    return map(groups) do g
+        chosen = representative(g)
+        return Dict{String,Any}(
+            "objective" => objective(chosen),
             "solves" => length(g),
             "configurations" => sort!(unique(["$(r["modeling"]) $(r["configuration"])" for r in g])),
-        ) for g in groups
-    ]
+            "reference" => reference(chosen),
+        )
+    end
 end
 
 const HEADER = """