diff --git a/benchmark/2_bench_state_estim.jl b/benchmark/2_bench_state_estim.jl
index 325e5ac3c..25fee68d2 100644
--- a/benchmark/2_bench_state_estim.jl
+++ b/benchmark/2_bench_state_estim.jl
@@ -272,18 +272,6 @@ mhe_cstr_daqp_pred = MovingHorizonEstimator(model; He, nint_u, σQint_u, optim,
 mhe_cstr_daqp_pred = setconstraint!(mhe_cstr_daqp_pred; v̂min, v̂max)
 JuMP.set_attribute(mhe_cstr_daqp_pred.optim, "eps_prox", 1e-6) # needed to support hessians H≥0
 
-optim = JuMP.Model(optimizer_with_attributes(Ipopt.Optimizer,"sb"=>"yes"), add_bridges=false)
-direct = true
-mhe_cstr_ipopt_curr = MovingHorizonEstimator(model; He, nint_u, σQint_u, optim, direct)
-mhe_cstr_ipopt_curr = setconstraint!(mhe_cstr_ipopt_curr; v̂min, v̂max)
-JuMP.unset_time_limit_sec(mhe_cstr_ipopt_curr.optim)
-
-optim = JuMP.Model(optimizer_with_attributes(Ipopt.Optimizer,"sb"=>"yes"), add_bridges=false)
-direct = false
-mhe_cstr_ipopt_pred = MovingHorizonEstimator(model; He, nint_u, σQint_u, optim, direct)
-mhe_cstr_ipopt_pred = setconstraint!(mhe_cstr_ipopt_pred; v̂min, v̂max)
-JuMP.unset_time_limit_sec(mhe_cstr_ipopt_pred.optim)
-
 samples, evals = 10000, 1
 CASE_ESTIM["CSTR"]["MovingHorizonEstimator"]["OSQP"]["Current form"] =
     @benchmarkable(test_mhe($mhe_cstr_osqp_curr, $plant); 
@@ -301,14 +289,6 @@ CASE_ESTIM["CSTR"]["MovingHorizonEstimator"]["DAQP"]["Prediction form"] =
     @benchmarkable(test_mhe($mhe_cstr_daqp_pred, $plant);
         samples=samples, evals=evals
     )
-CASE_ESTIM["CSTR"]["MovingHorizonEstimator"]["Ipopt"]["Current form"] =
-    @benchmarkable(test_mhe($mhe_cstr_ipopt_curr, $plant); 
-        samples=samples, evals=evals
-    )
-CASE_ESTIM["CSTR"]["MovingHorizonEstimator"]["Ipopt"]["Prediction form"] =
-    @benchmarkable(test_mhe($mhe_cstr_ipopt_pred, $plant);
-        samples=samples, evals=evals
-    )
 
 ## ---------------------- Case study: pendulum -------------------------------------------
 model, p = pendulum_model, pendulum_p
diff --git a/benchmark/3_bench_predictive_control.jl b/benchmark/3_bench_predictive_control.jl
index cee43de7d..ed4658bef 100644
--- a/benchmark/3_bench_predictive_control.jl
+++ b/benchmark/3_bench_predictive_control.jl
@@ -202,16 +202,6 @@ mpc_daqp_ss = setconstraint!(LinMPC(model; optim, transcription), ymin=[45, -Inf
 # mpc_daqp_ms = setconstraint!(LinMPC(model; optim, transcription), ymin=[45, -Inf])
 # JuMP.set_attribute(mpc_daqp_ms.optim, "eps_prox", 1e-6)
 
-optim = JuMP.Model(optimizer_with_attributes(Ipopt.Optimizer,"sb"=>"yes"), add_bridges=false)
-transcription = SingleShooting()
-mpc_ipopt_ss = setconstraint!(LinMPC(model; optim, transcription), ymin=[45, -Inf])
-JuMP.unset_time_limit_sec(mpc_ipopt_ss.optim)
-
-optim = JuMP.Model(optimizer_with_attributes(Ipopt.Optimizer,"sb"=>"yes"), add_bridges=false)
-transcription = MultipleShooting()
-mpc_ipopt_ms = setconstraint!(LinMPC(model; optim, transcription), ymin=[45, -Inf])
-JuMP.unset_time_limit_sec(mpc_ipopt_ms.optim) 
-
 samples, evals = 10000, 1
 CASE_MPC["CSTR"]["LinMPC"]["Without feedforward"]["OSQP"]["SingleShooting"] = 
     @benchmarkable(test_mpc($mpc_osqp_ss, $plant); 
@@ -225,14 +215,6 @@ CASE_MPC["CSTR"]["LinMPC"]["Without feedforward"]["DAQP"]["SingleShooting"] =
     @benchmarkable(test_mpc($mpc_daqp_ss, $plant); 
         samples=samples, evals=evals
     )
-CASE_MPC["CSTR"]["LinMPC"]["Without feedforward"]["Ipopt"]["SingleShooting"] =
-    @benchmarkable(test_mpc($mpc_ipopt_ss, $plant); 
-    samples=samples, evals=evals
-)
-CASE_MPC["CSTR"]["LinMPC"]["Without feedforward"]["Ipopt"]["MultipleShooting"] =
-    @benchmarkable(test_mpc($mpc_ipopt_ms, $plant); 
-        samples=samples, evals=evals
-    )
 
 ## ----------------- Case study: CSTR with feedforward -------------------------
 model_d = CSTR_model_d
@@ -271,16 +253,6 @@ mpc_d_daqp_ss = setconstraint!(LinMPC(model_d; optim, transcription), ymin=[45,
 # Skip DAQP with MultipleShooting, it is not designed for sparse Hessians. Kind of works 
 # with "eps_prox" configured to 1e-6, but not worth it.
 
-optim = JuMP.Model(optimizer_with_attributes(Ipopt.Optimizer,"sb"=>"yes"), add_bridges=false)
-transcription = SingleShooting()
-mpc_d_ipopt_ss = setconstraint!(LinMPC(model_d; optim, transcription), ymin=[45, -Inf])
-JuMP.unset_time_limit_sec(mpc_d_ipopt_ss.optim)
-
-optim = JuMP.Model(optimizer_with_attributes(Ipopt.Optimizer,"sb"=>"yes"), add_bridges=false)
-transcription = MultipleShooting()
-mpc_d_ipopt_ms = setconstraint!(LinMPC(model_d; optim, transcription), ymin=[45, -Inf])
-JuMP.unset_time_limit_sec(mpc_d_ipopt_ms.optim)
-
 samples, evals = 10000, 1
 CASE_MPC["CSTR"]["LinMPC"]["With feedforward"]["OSQP"]["SingleShooting"] = 
     @benchmarkable(test_mpc_d($mpc_d_osqp_ss, $plant); 
@@ -294,14 +266,6 @@ CASE_MPC["CSTR"]["LinMPC"]["With feedforward"]["DAQP"]["SingleShooting"] =
     @benchmarkable(test_mpc_d($mpc_d_daqp_ss, $plant); 
         samples=samples, evals=evals
     )
-CASE_MPC["CSTR"]["LinMPC"]["With feedforward"]["Ipopt"]["SingleShooting"] =
-    @benchmarkable(test_mpc_d($mpc_d_ipopt_ss, $plant); 
-        samples=samples, evals=evals
-    )
-CASE_MPC["CSTR"]["LinMPC"]["With feedforward"]["Ipopt"]["MultipleShooting"] =
-    @benchmarkable(test_mpc_d($mpc_d_ipopt_ms, $plant); 
-        samples=samples, evals=evals
-    )
 
 # ----------------- Case study: Pendulum noneconomic -----------------------------
 model, p = pendulum_model, pendulum_p
@@ -727,18 +691,6 @@ mpc3_daqp_ss = setconstraint!(mpc3_daqp_ss; umin, umax)
 # skip DAQP with MultipleShooting, it is not designed for sparse Hessians
 # did not found any settings that works well here (always reach the iteration limit).
 
-optim = JuMP.Model(optimizer_with_attributes(Ipopt.Optimizer,"sb"=>"yes"), add_bridges=false)
-transcription = SingleShooting()
-mpc3_ipopt_ss = LinMPC(kf; Hp, Hc, Mwt, Nwt, Cwt, optim, transcription)
-mpc3_ipopt_ss = setconstraint!(mpc3_ipopt_ss; umin, umax)
-JuMP.unset_time_limit_sec(mpc3_ipopt_ss.optim)
-
-optim = JuMP.Model(optimizer_with_attributes(Ipopt.Optimizer,"sb"=>"yes"), add_bridges=false)
-transcription = MultipleShooting()
-mpc3_ipopt_ms = LinMPC(kf; Hp, Hc, Mwt, Nwt, Cwt, optim, transcription)
-mpc3_ipopt_ms = setconstraint!(mpc3_ipopt_ms; umin, umax)
-JuMP.unset_time_limit_sec(mpc3_ipopt_ms.optim)
-
 samples, evals = 10000, 1
 CASE_MPC["Pendulum"]["LinMPC"]["Successive linearization"]["OSQP"]["SingleShooting"] = 
     @benchmarkable(
@@ -755,13 +707,3 @@ CASE_MPC["Pendulum"]["LinMPC"]["Successive linearization"]["DAQP"]["SingleShooti
         sim2!($mpc3_daqp_ss, $model, $N, $ry, $plant, $x_0, $x̂_0, $y_step),
         samples=samples, evals=evals
     )
-CASE_MPC["Pendulum"]["LinMPC"]["Successive linearization"]["Ipopt"]["SingleShooting"] = 
-    @benchmarkable(
-        sim2!($mpc3_ipopt_ss, $model, $N, $ry, $plant, $x_0, $x̂_0, $y_step),
-        samples=samples, evals=evals
-    )
-CASE_MPC["Pendulum"]["LinMPC"]["Successive linearization"]["Ipopt"]["MultipleShooting"] = 
-    @benchmarkable(
-        sim2!($mpc3_ipopt_ms, $model, $N, $ ry, $plant, $x_0, $x̂_0, $y_step),
-        samples=samples, evals=evals
-    )