Fix some issues and error handling fix

app.py

@@ -32,6 +32,11 @@ models_data = {
     "llama-3.3-70b-versatile": {"rpm": 30, "rpd": 1000, "tpm": 12000, "tpd": 100000},
     "llama3-70b-8192": {"rpm": 30, "rpd": 14400, "tpm": 6000, "tpd": 500000},
     "llama3-8b-8192": {"rpm": 30, "rpd": 14400, "tpm": 6000, "tpd": 500000},
+    "meta-llama/llama-4-maverick-17b-128e-instruct": {"rpm": 30, "rpd": 1000, "tpm": 6000, "tpd": None},
+    "meta-llama/llama-4-scout-17b-16e-instruct": {"rpm": 30, "rpd": 1000, "tpm": 30000, "tpd": None},
+    "meta-llama/llama-guard-4-12b": {"rpm": 30, "rpd": 14400, "tpm": 15000, "tpd": 500000},
+    "meta-llama/llama-prompt-guard-2-22m": {"rpm": 30, "rpd": 14400, "tpm": 15000, "tpd": None},
+    "meta-llama/llama-prompt-guard-2-86m": {"rpm": 30, "rpd": 14400, "tpm": None, "tpd": None},
 }
 
 model_list = [
@@ -121,12 +126,12 @@ def main_page():
 @app.post("/chat")
 def chat_with_groq(req: ChatRequest):
     models = req.models
-    if len(models) == 1 and models[0] == "":
+    if len(models) == 1 and (models[0] == "" or models[0] not in models_data.keys()):
         raise HTTPException(400, detail="Empty model field")
     messages = [clean_message(m) for m in req.messages]
     if len(models) == 1:
-        resp = router.completion(model=models[0], messages=messages, **req.model_dump(exclude={"models", "messages"}, exclude_defaults=True, exclude_none=True))
         try:
+            resp = router.completion(model=models[0], messages=messages, **req.model_dump(exclude={"models", "messages"}, exclude_defaults=True, exclude_none=True))
             print("Asked to", models[0], ":", messages)
             return {"error": False, "content": resp.choices[0].message.content}
         except Exception as e:
@@ -134,11 +139,14 @@ def chat_with_groq(req: ChatRequest):
             return {"error": True, "content": "Aucune clé ne fonctionne avec le modèle sélectionné, patientez ...."}
     else:
         for model in models:
-            resp = router.completion(model=model, messages=messages, **req.model_dump(exclude={"models", "messages"}, exclude_defaults=True, exclude_none=True))
+            if model not in models_data.keys():
+                print(f"Erreur: {model} n'existe pas")
+                continue
             try:
+                resp = router.completion(model=model, messages=messages, **req.model_dump(exclude={"models", "messages"}, exclude_defaults=True, exclude_none=True))
                 print("Asked to", models[0], ":", messages)
                 return {"error": False, "content": resp.choices[0].message.content}
             except Exception as e:
                 traceback.print_exception(e)
                 continue
-        return {"error": True, "content": "Aucune clé ne fonctionne avec le modèle sélectionné, patientez ...."}
+        return {"error": True, "content": "Tous les modèles n'ont pas fonctionné avec les différentes clé, patientez ...."}