diff --git a/src/model/model_evaluator.cpp b/src/model/model_evaluator.cpp
index 07ee125f8..eb7131406 100644
--- a/src/model/model_evaluator.cpp
+++ b/src/model/model_evaluator.cpp
@@ -53,6 +53,7 @@ struct evaluator_cfg : public default_rewriter_cfg {
     bool                            m_model_completion;
     bool                            m_cache;
     bool                            m_array_equalities;
+    bool                            m_array_as_stores;
 
     evaluator_cfg(ast_manager & m, model_core & md, params_ref const & p):
         m(m),
@@ -87,6 +88,7 @@ struct evaluator_cfg : public default_rewriter_cfg {
         m_model_completion = p.completion();
         m_cache            = p.cache();
         m_array_equalities = p.array_equalities();
+        m_array_as_stores  = p.array_as_stores();
     }
 
     bool evaluate(func_decl * f, unsigned num, expr * const * args, expr_ref & result) {
@@ -196,20 +198,22 @@ struct evaluator_cfg : public default_rewriter_cfg {
             if (evaluate(a->get_decl(), a->get_num_args(), a->get_args(), result)) {
                 return BR_REWRITE1;
             }
+            if (false && m_array_as_stores && m_ar.is_array(result)) {
+                expand_stores(result);
+            }
         }
         CTRACE("model_evaluator", st != BR_FAILED, tout << result << "\n";);
         return st;
     }
 
-    void expand_value(expr_ref& val) {
+    void expand_stores(expr_ref& val) {
         vector<expr_ref_vector> stores;
         expr_ref else_case(m);
         bool _unused;
         if (m_ar.is_array(val) && extract_array_func_interp(val, stores, else_case, _unused)) {
             sort* srt = m.get_sort(val);
             val = m_ar.mk_const_array(srt, else_case);
-            for (unsigned i = stores.size(); i > 0; ) {
-                --i;
+            for (unsigned i = stores.size(); i-- > 0; ) {
                 expr_ref_vector args(m);
                 args.push_back(val);
                 args.append(stores[i].size(), stores[i].c_ptr());
@@ -288,7 +292,6 @@ struct evaluator_cfg : public default_rewriter_cfg {
 
     bool cache_results() const { return m_cache; }
 
-
     br_status mk_array_eq(expr* a, expr* b, expr_ref& result) {
         if (a == b) {
             result = m.mk_true();
@@ -497,9 +500,6 @@ struct evaluator_cfg : public default_rewriter_cfg {
         TRACE("model_evaluator", tout << "else case: " << mk_pp(else_case, m) << "\n";);
         return true;
     }
-
-
-
 };
 
 template class rewriter_tpl<evaluator_cfg>;
@@ -513,10 +513,7 @@ struct model_evaluator::imp : public rewriter_tpl<evaluator_cfg> {
         m_cfg(md.get_manager(), md, p) {
         set_cancel_check(false);
     }
-
-    void expand_value (expr_ref &val) {
-        m_cfg.expand_value (val);
-    }
+    void expand_stores(expr_ref &val) {m_cfg.expand_stores(val);}
 };
 
 model_evaluator::model_evaluator(model_core & md, params_ref const & p) {
@@ -571,7 +568,7 @@ void model_evaluator::reset(model_core &model, params_ref const& p) {
 void model_evaluator::operator()(expr * t, expr_ref & result) {
     TRACE("model_evaluator", tout << mk_ismt2_pp(t, m()) << "\n";);
     m_imp->operator()(t, result);
-    m_imp->expand_value(result);
+    m_imp->expand_stores(result);
 }
 
 expr_ref model_evaluator::operator()(expr * t) {
@@ -581,6 +578,13 @@ expr_ref model_evaluator::operator()(expr * t) {
     return result;
 }
 
+expr_ref_vector model_evaluator::operator()(expr_ref_vector const& ts) {
+    expr_ref_vector rs(m());
+    for (expr* t : ts) rs.push_back((*this)(t));
+    return rs;
+}
+
+
 bool model_evaluator::is_true(expr* t) {
     expr_ref tmp(m());
     return eval(t, tmp, true) && m().is_true(tmp);
@@ -601,12 +605,12 @@ bool model_evaluator::eval(expr* t, expr_ref& r, bool model_completion) {
     try {
         r = (*this)(t);
         return true;
-    } 
+    }
     catch (model_evaluator_exception &ex) {
         (void)ex;
         TRACE("model_evaluator", tout << ex.msg () << "\n";);
         return false;
-    }    
+    }
 }
 
 bool model_evaluator::eval(expr_ref_vector const& ts, expr_ref& r, bool model_completion) {
@@ -614,6 +618,3 @@ bool model_evaluator::eval(expr_ref_vector const& ts, expr_ref& r, bool model_co
     tmp = mk_and(ts);
     return eval(tmp, r, model_completion);
 }
-
-
-
diff --git a/src/model/model_evaluator.h b/src/model/model_evaluator.h
index 6ae7d8891..d9bf3c375 100644
--- a/src/model/model_evaluator.h
+++ b/src/model/model_evaluator.h
@@ -44,6 +44,7 @@ public:
 
     void operator()(expr * t, expr_ref & r);
     expr_ref operator()(expr* t);
+    expr_ref_vector operator()(expr_ref_vector const& ts);
 
     // exception safe
     bool eval(expr* t, expr_ref& r, bool model_completion = true);
@@ -53,6 +54,11 @@ public:
     bool is_false(expr * t);
     bool is_true(expr_ref_vector const& ts);
 
+    /**
+     * best effort evaluator of extensional array equality.
+     */
+    expr_ref eval_array_eq(app* e, expr* arg1, expr* arg2);
+
     void cleanup(params_ref const & p = params_ref());
     void reset(params_ref const & p = params_ref());
     void reset(model_core& model, params_ref const & p = params_ref());
diff --git a/src/model/model_evaluator_params.pyg b/src/model/model_evaluator_params.pyg
index b6417f7fc..509b3e7c7 100644
--- a/src/model/model_evaluator_params.pyg
+++ b/src/model/model_evaluator_params.pyg
@@ -4,6 +4,7 @@ def_module_params('model_evaluator',
                           max_steps_param(),
                           ('completion', BOOL, False, 'assigns an interptetation to symbols that do not have one in the current model, when evaluating expressions in the current model'),
                           ('cache', BOOL, True, 'cache intermediate results in the model evaluator'),
-                          ('array_equalities', BOOL, True, 'evaluate array equalities')
+                          ('array_equalities', BOOL, True, 'evaluate array equalities'),
+                          ('array_as_stores', BOOL, True, 'return array as a set of stores'),
                           ))