diff --git a/Project.toml b/Project.toml
index a9fbf620..a8fb99ce 100644
--- a/Project.toml
+++ b/Project.toml
@@ -36,12 +36,12 @@ IterativeSolvers = "~0.9.1"
 JuMP = "~0.21"
 LightGraphs = "~1.3"
 MKL_jll = "~2020"
-MUMPS_seq_jll = "~5.2.1, ~5.3.5"
+MUMPS_seq_jll = "~5.2.1, 5.3.5"
 MathOptInterface = "~0.9"
 Metis = "~1.0"
 NLPModels = "~0.14"
 OpenBLAS32_jll = "~0.3"
-Plasmo = "~0.3"
+Plasmo = "~0.3, 0.4"
 SolverCore = "~0.1"
 StaticArrays = "~0.12, 1.0"
 julia = "1.6"
diff --git a/src/Interfaces/Plasmo_interface.jl b/src/Interfaces/Plasmo_interface.jl
index fe81529e..ee2f90b3 100644
--- a/src/Interfaces/Plasmo_interface.jl
+++ b/src/Interfaces/Plasmo_interface.jl
@@ -4,7 +4,15 @@
 const dummy_function = ()->nothing
 
 num_linkconstraints(modeledge::OptiEdge) = length(modeledge.linkconstraints)
-moi_optimizer(modelnode::OptiNode) = modelnode.model.moi_backend.optimizer.model
+function _caching_optimizer(modelnode::OptiNode)
+    if isa(modelnode.model.moi_backend,MOIU.CachingOptimizer)
+        return modelnode.model.moi_backend
+    else
+        return modelnode.model.moi_backend.optimizer
+    end
+end
+moi_optimizer(modelnode::OptiNode) = _caching_optimizer(modelnode).optimizer.model
+_caching_optimizer(model::Any) = model.moi_backend
 function set_g_link!(linkedge::OptiEdge,l,gl,gu)
     cnt = 1
     for (ind,linkcon) in linkedge.linkconstraints
@@ -165,9 +173,9 @@ function NonlinearProgram(graph::OptiGraph)
     for modelnode in modelnodes
         num_variables(modelnode) == 0 && error("Empty node exist! Delete the empty nodes.")
     end
-    
+
     @blas_safe_threads for k=1:length(modelnodes)
-        set_optimizer(modelnodes[k].model,Optimizer)
+        set_optimizer(modelnodes[k],Optimizer)
         if modelnodes[k].model.nlp_data !== nothing
             MOI.set(modelnodes[k].model, MOI.NLPBlock(),
                     _create_nlp_block_data(modelnodes[k].model))
@@ -234,7 +242,8 @@ function NonlinearProgram(graph::OptiGraph)
 
     modelmap=Dict(modelnodes[k].model=> k for k=1:K)
     x_index_map = Dict(
-        var=>ninds[modelmap[var.model]][backend(var.model).model_to_optimizer_map[var.index].value]
+        # var=>ninds[modelmap[var.model]][backend(var.model).optimizer.model_to_optimizer_map[var.index].value]
+        var=>ninds[modelmap[var.model]][_caching_optimizer(getnode(var)).model_to_optimizer_map[var.index].value]
         for modelnode in modelnodes for var in all_variables(modelnode))
     cnt = 0
     g_index_map = Dict(con=> m + (cnt+=1) for linkedge in linkedges for (ind,con) in linkedge.linkconstraints)
diff --git a/src/MadNLP.jl b/src/MadNLP.jl
index 0391e0f1..b6ccc267 100644
--- a/src/MadNLP.jl
+++ b/src/MadNLP.jl
@@ -18,7 +18,7 @@ import Base: string, show, print, size, getindex, copyto!, @kwdef
 import StaticArrays: SVector, setindex
 import SuiteSparse: UMFPACK
 import LightGraphs: Graph, Edge, add_edge!, edges, src, dst, neighbors, nv
-import Plasmo: OptiGraph, OptiNode, OptiEdge, all_nodes, all_edges, all_variables, num_all_nodes, num_variables, getlinkconstraints
+import Plasmo: OptiGraph, OptiNode, OptiEdge, all_nodes, all_edges, all_variables, num_all_nodes, num_variables, getlinkconstraints, getnode
 import JuMP: _create_nlp_block_data, set_optimizer, GenericAffExpr, backend, termination_status
 import NLPModels: finalize, AbstractNLPModel, obj, grad!, cons!, jac_coord!, hess_coord!, hess_structure!, jac_structure!
 import SolverCore: GenericExecutionStats