diff --git a/apps/graph/cartesian_function_store.cpp b/apps/graph/cartesian_function_store.cpp
index 341ead7be..98af7af7c 100644
--- a/apps/graph/cartesian_function_store.cpp
+++ b/apps/graph/cartesian_function_store.cpp
@@ -15,9 +15,10 @@ Ion::Storage::Record::ErrorStatus CartesianFunctionStore::addEmptyModel() {
   return error;
 }
 
-void CartesianFunctionStore::setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record record) const {
+ExpressionModelHandle * CartesianFunctionStore::setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record record) const {
   assert(cacheIndex >= 0 && cacheIndex < maxNumberOfMemoizedModels());
   m_functions[cacheIndex] = CartesianFunction(record);
+  return &m_functions[cacheIndex];
 }
 
 ExpressionModelHandle * CartesianFunctionStore::memoizedModelAtIndex(int cacheIndex) const {
diff --git a/apps/graph/cartesian_function_store.h b/apps/graph/cartesian_function_store.h
index f2ff11601..365de72ab 100644
--- a/apps/graph/cartesian_function_store.h
+++ b/apps/graph/cartesian_function_store.h
@@ -16,7 +16,7 @@ public:
 private:
   Ion::Storage::Record::ErrorStatus addEmptyModel() override;
   const char * modelExtension() const override { return Ion::Storage::funcExtension; }
-  void setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record record) const override;
+  Shared::ExpressionModelHandle * setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record record) const override;
   Shared::ExpressionModelHandle * memoizedModelAtIndex(int cacheIndex) const override;
   mutable Shared::CartesianFunction m_functions[k_maxNumberOfMemoizedModels];
 };
diff --git a/apps/sequence/sequence_store.cpp b/apps/sequence/sequence_store.cpp
index 6f9ca7dfa..5e70ee688 100644
--- a/apps/sequence/sequence_store.cpp
+++ b/apps/sequence/sequence_store.cpp
@@ -47,9 +47,14 @@ int SequenceStore::sequenceIndexForName(char name) {
   return 0;
 }
 
-void SequenceStore::setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record record) const {
+Shared::ExpressionModelHandle * SequenceStore::setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record record) const {
   assert(cacheIndex >= 0 && cacheIndex < maxNumberOfMemoizedModels());
-  m_sequences[cacheIndex] = Sequence(record);
+  int index = cacheIndex;
+  if (!record.isNull()) {
+    index = SequenceStore::sequenceIndexForName(record.fullName()[0]);
+  }
+  m_sequences[index] = Sequence(record);
+  return &m_sequences[index];
 }
 
 Shared::ExpressionModelHandle * SequenceStore::memoizedModelAtIndex(int cacheIndex) const {
diff --git a/apps/sequence/sequence_store.h b/apps/sequence/sequence_store.h
index 9528fd419..5f5a27a4b 100644
--- a/apps/sequence/sequence_store.h
+++ b/apps/sequence/sequence_store.h
@@ -32,9 +32,16 @@ public:
 
 private:
   const char * modelExtension() const override { return Ion::Storage::seqExtension; }
-  /* We don't really use model memoization as the number of Sequence is limited
-   * and we keep enough Sequences to store them all. */
-  void setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record record) const override;
+  /* We don't use model memoization for two reasons:
+   * - the number of Sequence is capped so we keep enough Sequences to store them all.
+   * - evaluating sequences require to evaluate all of them at the same time
+   *   (as they might be co-dependent) which doesn't suit our ring memoization.
+   * To still make it work with the ExpressionModelStore, we force
+   * setMemoizedModelAtIndex to store u, v and w sequences in this order in
+   * m_sequences whatever the value of cacheIndex. We thus return a
+   * ExpressionModelHandle pointer after setting the model as it won't be the
+   * memoized model at cacheIndex. */
+  Shared::ExpressionModelHandle * setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record record) const override;
   Shared::ExpressionModelHandle * memoizedModelAtIndex(int cacheIndex) const override;
   mutable Sequence m_sequences[MaxNumberOfSequences];
 };
diff --git a/apps/shared/expression_model_store.cpp b/apps/shared/expression_model_store.cpp
index eee7cb969..8ba25529c 100644
--- a/apps/shared/expression_model_store.cpp
+++ b/apps/shared/expression_model_store.cpp
@@ -21,8 +21,7 @@ ExpressionModelHandle * ExpressionModelStore::privateModelForRecord(Ion::Storage
       return memoizedModelAtIndex(i);
     }
   }
-  setMemoizedModelAtIndex(m_oldestMemoizedIndex, record);
-  ExpressionModelHandle * result = memoizedModelAtIndex(m_oldestMemoizedIndex);
+  ExpressionModelHandle * result = setMemoizedModelAtIndex(m_oldestMemoizedIndex, record);
   m_oldestMemoizedIndex = (m_oldestMemoizedIndex+1) % maxNumberOfMemoizedModels();
   return result;
 }
diff --git a/apps/shared/expression_model_store.h b/apps/shared/expression_model_store.h
index 31b93dd3f..2cf45424d 100644
--- a/apps/shared/expression_model_store.h
+++ b/apps/shared/expression_model_store.h
@@ -43,7 +43,7 @@ protected:
   ExpressionModelHandle * privateModelForRecord(Ion::Storage::Record record) const;
 private:
   void resetMemoizedModelsExceptRecord(const Ion::Storage::Record record = Ion::Storage::Record()) const;
-  virtual void setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record) const = 0;
+  virtual ExpressionModelHandle * setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record) const = 0;
   virtual ExpressionModelHandle * memoizedModelAtIndex(int cacheIndex) const = 0;
   virtual const char * modelExtension() const = 0;
   /* Memoization of k_maxNumberOfMemoizedModels. When the required model is not
diff --git a/apps/solver/equation_store.cpp b/apps/solver/equation_store.cpp
index 9f2b75480..d083bb533 100644
--- a/apps/solver/equation_store.cpp
+++ b/apps/solver/equation_store.cpp
@@ -44,9 +44,10 @@ Ion::Storage::Record::ErrorStatus EquationStore::addEmptyModel() {
   return Ion::Storage::sharedStorage()->createRecordWithExtension(name, Ion::Storage::eqExtension, nullptr, 0);
 }
 
-void EquationStore::setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record record) const {
+Shared::ExpressionModelHandle * EquationStore::setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record record) const {
   assert(cacheIndex >= 0 && cacheIndex < maxNumberOfMemoizedModels());
   m_equations[cacheIndex] = Equation(record);
+  return &m_equations[cacheIndex];
 }
 
 ExpressionModelHandle * EquationStore::memoizedModelAtIndex(int cacheIndex) const {
diff --git a/apps/solver/equation_store.h b/apps/solver/equation_store.h
index 70737f088..b1a1bf717 100644
--- a/apps/solver/equation_store.h
+++ b/apps/solver/equation_store.h
@@ -81,7 +81,7 @@ private:
   const char * modelExtension() const override { return Ion::Storage::eqExtension; }
   /* We don't really use model memoization as the number of Equation is limited
    * and we keep enough Equations to store them all. */
-  void setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record record) const override;
+  Shared::ExpressionModelHandle * setMemoizedModelAtIndex(int cacheIndex, Ion::Storage::Record record) const override;
   Shared::ExpressionModelHandle * memoizedModelAtIndex(int cacheIndex) const override;
 
   Error resolveLinearSystem(Poincare::Expression solutions[k_maxNumberOfExactSolutions], Poincare::Expression solutionApproximations[k_maxNumberOfExactSolutions], Poincare::Expression coefficients[k_maxNumberOfEquations][Poincare::Expression::k_maxNumberOfVariables], Poincare::Expression constants[k_maxNumberOfEquations], Poincare::Context * context);