clean error messages, add model validation, add gemini-3.1-flash-lite

.graphify_uncached.txt

@@ -0,0 +1,15 @@
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\config.py
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\main.py
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\.opencode\opencode.json
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\.opencode\package.json
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\.opencode\plugins\graphify.js
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\proxy\auth.py
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\proxy\client.py
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\proxy\providers.py
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\proxy\router.py
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\proxy\__init__.py
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\DESIGN.md
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\PRODUCT.md
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\README.md
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\requirements.txt
+C:\Users\Rafandra Aydin\Documents\Wimmboo\OpenAIProxy\.opencode\AGENTS.md

proxy/client.py

@@ -52,6 +52,22 @@ async def _poll_nvidia(client: httpx.AsyncClient, provider: Provider, request_id
     raise HTTPException(status_code=504, detail=f"NVIDIA async job {request_id} timed out after polling")
 
 
+def _upstream_error_message(status_code: int, content: bytes) -> str:
+    """Try to extract a readable message from a provider error response."""
+    text = content.decode("utf-8", errors="replace").strip()
+    if not text:
+        return f"Upstream provider returned HTTP {status_code}"
+    try:
+        data = json.loads(text)
+        if isinstance(data, dict):
+            if isinstance(data.get("error"), dict):
+                return data["error"].get("message") or json.dumps(data["error"])
+            return data.get("message") or data.get("error") or text
+    except Exception:
+        pass
+    return text
+
+
 async def _stream_response(provider: Provider, real_key: str, body: dict[str, Any]) -> AsyncIterator[bytes]:
     """Async generator that manages its own httpx client for streaming."""
     target_url = f"{provider.base_url}/v1/chat/completions"
@@ -60,7 +76,8 @@ async def _stream_response(provider: Provider, real_key: str, body: dict[str, An
         async with client.stream("POST", target_url, headers=headers, json=body) as upstream:
             if upstream.status_code >= 400:
                 content = await upstream.aread()
-                raise HTTPException(status_code=upstream.status_code, detail=content.decode("utf-8", errors="replace"))
+                detail = _upstream_error_message(upstream.status_code, content)
+                raise HTTPException(status_code=upstream.status_code, detail=detail)
             async for chunk in upstream.aiter_raw():
                 yield chunk
 
@@ -91,7 +108,8 @@ async def forward_request(provider: Provider, real_key: str, request: Request, b
         if upstream.status_code >= 400:
             content = await upstream.aread()
             logger.warning("Upstream error from %s: %d %s", provider.name, upstream.status_code, content[:500])
-            raise HTTPException(status_code=upstream.status_code, detail=content.decode("utf-8", errors="replace"))
+            detail = _upstream_error_message(upstream.status_code, content)
+            raise HTTPException(status_code=upstream.status_code, detail=detail)
 
         return {
             "status_code": upstream.status_code,

proxy/router.py

@@ -24,10 +24,11 @@ MODELS = {
         "moonshotai/kimi-k2.5",
         "google/gemma-4-31b-it",
         "google/gemma-4-26b-it",
-        "google/gemma-3-27b-it",
     ],
     "google": [
+        "gemini-3.1-flash-lite",
         "gemini-2.5-pro-001",
+        "gemini-2.5-flash-001",
         "gemini-2.0-flash-001",
         "gemini-2.0-flash-lite-001",
         "gemini-1.5-pro-001",
@@ -66,7 +67,15 @@ async def chat_completions(request: Request) -> Response:
     if not isinstance(body, dict):
         raise StarletteHTTPException(status_code=400, detail="Request body must be a JSON object")
 
-    logger.info("Proxying request to provider=%s model=%s", provider.name, body.get("model", "default"))
+    model = body.get("model")
+    allowed = MODELS.get(provider.name, [])
+    if model and model not in allowed:
+        raise StarletteHTTPException(
+            status_code=400,
+            detail=f"Model '{model}' is not available for provider '{provider.name}'. Available: {', '.join(allowed)}",
+        )
+
+    logger.info("Proxying request to provider=%s model=%s", provider.name, model or "default")
 
     result = await forward_request(provider, real_key, request, body)