diff --git a/libraries/gpu/src/gpu/GLBackend.cpp b/libraries/gpu/src/gpu/GLBackend.cpp
index 08a9a39d68..f337d5c3e0 100644
--- a/libraries/gpu/src/gpu/GLBackend.cpp
+++ b/libraries/gpu/src/gpu/GLBackend.cpp
@@ -96,8 +96,11 @@ void GLBackend::render(Batch& batch) {
     }
 }
 
-void GLBackend::renderBatch(Batch& batch) {
+void GLBackend::renderBatch(Batch& batch, bool syncCache) {
     GLBackend backend;
+    if (syncCache) {
+        backend.syncCache();
+    }
     backend.render(batch);
 }
 
@@ -134,6 +137,12 @@ bool GLBackend::checkGLError(const char* name) {
     }
 }
 
+
+void GLBackend::syncCache() {
+    syncTransformStateCache();
+    syncPipelineStateCache();
+}
+
 void GLBackend::do_draw(Batch& batch, uint32 paramOffset) {
     updateInput();
     updateTransform();
@@ -547,4 +556,3 @@ void GLBackend::fetchMatrix(GLenum target, glm::mat4 & m) {
 }
 
 
-
diff --git a/libraries/gpu/src/gpu/GLBackend.h b/libraries/gpu/src/gpu/GLBackend.h
index 3551953998..7546484cc3 100644
--- a/libraries/gpu/src/gpu/GLBackend.h
+++ b/libraries/gpu/src/gpu/GLBackend.h
@@ -24,12 +24,20 @@ namespace gpu {
 class GLBackend : public Backend {
 public:
 
+    explicit GLBackend(bool syncCache);
     GLBackend();
     ~GLBackend();
 
     void render(Batch& batch);
 
-    static void renderBatch(Batch& batch);
+    // Render Batch create a local Context and execute the batch with it
+    // WARNING:
+    // if syncCache is true, then the gpu::GLBackend will synchornize
+    // its cache with the current gl state and it's BAD
+    // If you know you don't rely on any state changed by naked gl calls then
+    // leave to false where it belongs
+    // if true, the needed resync IS EXPENSIVE
+    static void renderBatch(Batch& batch, bool syncCache = true);
 
     static bool checkGLError(const char* name = nullptr);
 
@@ -79,6 +87,7 @@ public:
     static GLShader* syncGPUObject(const Shader& shader);
     static GLuint getShaderID(const ShaderPointer& shader);
 
+    // FIXME: Please remove these 2 calls once the text renderer doesn't use naked gl calls anymore
     static void loadMatrix(GLenum target, const glm::mat4 & m);
     static void fetchMatrix(GLenum target, glm::mat4 & m);
 
@@ -186,6 +195,12 @@ public:
 
     void do_setStateColorWriteMask(uint32 mask);
 
+    // This call synchronize the Full Backend cache with the current GLState
+    // THis is only intended to be used when mixing raw gl calls with the gpu api usage in order to sync
+    // the gpu::Backend state with the true gl state which has probably been messed up by these ugly naked gl calls
+    // Let's try to avoid to do that as much as possible!
+    void syncCache();
+
 protected:
 
     // Draw Stage
@@ -241,6 +256,8 @@ protected:
 
     void initTransform();
     void killTransform();
+    // Synchronize the state cache of this Backend with the actual real state of the GL Context
+    void syncTransformStateCache();
     void updateTransform();
     struct TransformStageState {
         TransformObject _transformObject;
@@ -297,7 +314,7 @@ protected:
 
         GLState* _state;
         bool _invalidState = false;
-        bool _needStateSync = true;
+      //  bool _needStateSync = true;
 
         PipelineStageState() :
             _pipeline(),
@@ -306,8 +323,8 @@ protected:
             _stateCache(State::DEFAULT),
             _stateSignatureCache(0),
             _state(nullptr),
-            _invalidState(false),
-            _needStateSync(true)
+            _invalidState(false)//,
+      //      _needStateSync(true)
              {}
     } _pipeline;
 
diff --git a/libraries/gpu/src/gpu/GLBackendPipeline.cpp b/libraries/gpu/src/gpu/GLBackendPipeline.cpp
index 938ed77730..f124770cfb 100755
--- a/libraries/gpu/src/gpu/GLBackendPipeline.cpp
+++ b/libraries/gpu/src/gpu/GLBackendPipeline.cpp
@@ -64,10 +64,10 @@ void GLBackend::do_setPipeline(Batch& batch, uint32 paramOffset) {
         return;
     }
    
-    if (_pipeline._needStateSync) {
+   /* if (_pipeline._needStateSync) {
         syncPipelineStateCache();
         _pipeline._needStateSync = false;
-    }
+    }*/
 
     // null pipeline == reset
     if (!pipeline) {
@@ -111,12 +111,12 @@ void GLBackend::do_setPipeline(Batch& batch, uint32 paramOffset) {
 #define DEBUG_GLSTATE
 void GLBackend::updatePipeline() {
 #ifdef DEBUG_GLSTATE
-    if (_pipeline._needStateSync) {
+  /*  if (_pipeline._needStateSync) {
          State::Data state;
          getCurrentGLState(state);
          State::Signature signature = State::evalSignature(state);
          (void) signature; // quiet compiler
-    }
+    }*/
 #endif
 
     if (_pipeline._invalidProgram) {
diff --git a/libraries/gpu/src/gpu/GLBackendTransform.cpp b/libraries/gpu/src/gpu/GLBackendTransform.cpp
index d619d0afee..3f760e4cc8 100755
--- a/libraries/gpu/src/gpu/GLBackendTransform.cpp
+++ b/libraries/gpu/src/gpu/GLBackendTransform.cpp
@@ -55,6 +55,25 @@ void GLBackend::killTransform() {
 #else
 #endif
 }
+
+void GLBackend::syncTransformStateCache() {
+    _transform._invalidProj = true;
+    _transform._invalidView = true;
+    _transform._invalidModel = true;
+
+    GLint currentMode;
+    glGetIntegerv(GL_MATRIX_MODE, &currentMode);
+    _transform._lastMode = currentMode;
+
+    glGetFloatv(GL_PROJECTION_MATRIX, (float*) &_transform._projection);
+
+    Mat4 modelView;
+    glGetFloatv(GL_MODELVIEW_MATRIX, (float*) &modelView);
+    auto modelViewInv = glm::inverse(modelView);
+    _transform._view.evalFromRawMatrix(modelViewInv);
+    _transform._model.setIdentity();
+}
+
 void GLBackend::updateTransform() {
     // Check all the dirty flags and update the state accordingly
     if (_transform._invalidProj) {
diff --git a/libraries/render-utils/src/Model.cpp b/libraries/render-utils/src/Model.cpp
index c51ca9682a..2d0c6ec735 100644
--- a/libraries/render-utils/src/Model.cpp
+++ b/libraries/render-utils/src/Model.cpp
@@ -935,8 +935,8 @@ bool Model::renderCore(float alpha, RenderMode mode, RenderArgs* args) {
             glPushMatrix();
         #endif
 
-        ::gpu::GLBackend::renderBatch(batch);
-
+        ::gpu::GLBackend::renderBatch(batch, true); // force sync with gl state here
+ 
         #if defined(ANDROID)
         #else
             glPopMatrix();
@@ -1846,6 +1846,7 @@ void Model::endScene(RenderMode mode, RenderArgs* args) {
     }
 
     gpu::GLBackend backend;
+    backend.syncCache(); // force sync with gl state here
 
     if (args) {
         glm::mat4 proj;