Update backend/model_router.py

backend/model_router.py

@@ -1,8 +1,5 @@
 import os, random, aiohttp
 from typing import List, Dict, AsyncGenerator
-
-# Correct imports for groq 0.3.0+
-from groq import Client as Groq
 from openai import AsyncOpenAI
 
 groq_keys = os.environ.get("GROQ_API_KEYS", "").split(",")
@@ -10,25 +7,29 @@ openrouter_keys = os.environ.get("OPENROUTER_KEYS", "").split(",")
 
 POLLINATIONS_URL = "https://text.pollinations.ai/"
 
+# Groq provides an OpenAI-compatible API – we use AsyncOpenAI directly.
+def get_groq_client():
+    key = random.choice(groq_keys)
+    return AsyncOpenAI(base_url="https://api.groq.com/openai/v1", api_key=key)
+
+def get_openrouter_client():
+    key = random.choice(openrouter_keys)
+    return AsyncOpenAI(base_url="https://openrouter.ai/api/v1", api_key=key)
+
 async def route_chat(model: str, messages: List[Dict]) -> AsyncGenerator[str, None]:
     if model.startswith("groq-"):
-        key = random.choice(groq_keys)
-        client = Groq(api_key=key)
+        client = get_groq_client()
         model_id = model[5:]   # strip "groq-"
-        stream = client.chat.completions.create(
+        stream = await client.chat.completions.create(
             messages=messages,
             model=model_id,
             stream=True,
         )
-        for chunk in stream:
+        async for chunk in stream:
             if chunk.choices[0].delta.content:
                 yield chunk.choices[0].delta.content
     elif model.startswith("openrouter-"):
-        key = random.choice(openrouter_keys)
-        client = AsyncOpenAI(
-            base_url="https://openrouter.ai/api/v1",
-            api_key=key,
-        )
+        client = get_openrouter_client()
         model_id = model[11:]  # strip "openrouter-"
         stream = await client.chat.completions.create(
             messages=messages,
@@ -49,17 +50,18 @@ async def route_chat(model: str, messages: List[Dict]) -> AsyncGenerator[str, No
         raise ValueError(f"Unknown model: {model}")
 
 async def route_autocomplete(model: str, prefix: str, suffix: str, max_tokens: int) -> str:
-    key = random.choice(groq_keys)
-    client = Groq(api_key=key)
-    prompt = f"<fim_prefix>{prefix}<fim_suffix>{suffix}<fim_middle>"
-    response = client.completions.create(
-        model="codellama-70b",
-        prompt=prompt,
+    # Use Groq's chat completion for autocomplete
+    client = get_groq_client()
+    prompt = f"Complete the following code:\n```\n{prefix}\n```"
+    if suffix:
+        prompt += f"\nThe code should continue until before: {suffix}"
+    response = await client.chat.completions.create(
+        model="codellama-70b-instruct",   # or any other model you prefer
+        messages=[{"role": "user", "content": prompt}],
         max_tokens=max_tokens,
         temperature=0.1,
-        stop=["<fim_middle>"],
     )
-    return response.choices[0].text
+    return response.choices[0].message.content
 
 def get_available_models():
     return [