Updated

app.py

@@ -1858,6 +1858,93 @@ def root():
     })
 
 
+@app.route('/debug/raw-test', methods=['GET'])
+def debug_raw_test():
+    """Test Onyx directly with and without llm_override"""
+    import time as t
+    results = []
+    
+    endpoints = [
+        f"{ONYX_BASE_URL}/api/chat/send-chat-message",
+        f"{ONYX_BASE_URL}/api/chat/send-message",
+    ]
+    
+    # Create session
+    sess = requests.post(
+        f"{ONYX_BASE_URL}/api/chat/create-chat-session",
+        json={"persona_id": 1, "description": "Debug Test"},
+        headers=get_headers(), timeout=15
+    )
+    if sess.status_code != 200:
+        return jsonify({"error": f"Session creation failed: {sess.text}"}), 500
+    sid = sess.json().get("chat_session_id") or sess.json().get("id")
+    
+    # Test 1: WITHOUT llm_override
+    payload1 = {
+        "message": "Hello, say hi back in 3 words.",
+        "chat_session_id": sid,
+        "stream": True
+    }
+    
+    test1_result = {"test": "WITHOUT llm_override", "payload": payload1}
+    for url in endpoints:
+        try:
+            r = requests.post(url, json=payload1, headers=get_headers(), stream=True, timeout=30)
+            if r.status_code == 404:
+                continue
+            chunks = []
+            for chunk in r.iter_content(chunk_size=None, decode_unicode=True):
+                if chunk:
+                    chunks.append(chunk[:200])
+            test1_result["status"] = r.status_code
+            test1_result["endpoint"] = url.split("/")[-1]
+            test1_result["raw_chunks"] = chunks[:10]
+            break
+        except Exception as e:
+            test1_result["error"] = str(e)
+    results.append(test1_result)
+    
+    # Create new session for test 2
+    sess2 = requests.post(
+        f"{ONYX_BASE_URL}/api/chat/create-chat-session",
+        json={"persona_id": 1, "description": "Debug Test 2"},
+        headers=get_headers(), timeout=15
+    )
+    sid2 = sess2.json().get("chat_session_id") or sess2.json().get("id")
+    
+    # Test 2: WITH llm_override
+    payload2 = {
+        "message": "Hello, say hi back in 3 words.",
+        "chat_session_id": sid2,
+        "stream": True,
+        "llm_override": {
+            "model_provider": "Anthropic",
+            "model_version": "claude-opus-4-5",
+            "temperature": 0.7
+        }
+    }
+    
+    test2_result = {"test": "WITH llm_override (claude-opus-4-5)", "payload": payload2}
+    for url in endpoints:
+        try:
+            r = requests.post(url, json=payload2, headers=get_headers(), stream=True, timeout=30)
+            if r.status_code == 404:
+                continue
+            chunks = []
+            for chunk in r.iter_content(chunk_size=None, decode_unicode=True):
+                if chunk:
+                    chunks.append(chunk[:200])
+            test2_result["status"] = r.status_code
+            test2_result["endpoint"] = url.split("/")[-1]
+            test2_result["raw_chunks"] = chunks[:10]
+            break
+        except Exception as e:
+            test2_result["error"] = str(e)
+    results.append(test2_result)
+    
+    return jsonify({"session_ids": [sid, sid2], "results": results})
+
+
 # ============== Error Handlers ==============
 
 @app.errorhandler(404)