Task/refactor 2 (#37)

fix responses endpoint

Author: OriNachum

src/open_responses_server/api_controller.py

@@ -403,7 +403,7 @@ async def stream_proxy():
                 async with client.stream(request.method, url, headers=headers, content=body, timeout=120.0) as response:
                     async for chunk in response.aiter_bytes():
                         yield chunk
-            return StreamingResponse(stream_proxy(), media_type=response.headers.get("content-type"))
+            return StreamingResponse(stream_proxy(), media_type=request.headers.get('accept', 'application/json'))
         else:
             response = await client.request(request.method, url, headers=headers, content=body, timeout=120.0)
             return Response(content=response.content, status_code=response.status_code, headers=response.headers)

test_tools/config.sh

@@ -3,9 +3,9 @@
 # This file should be sourced by other scripts
 
 # Default model ID to use if DEFAULT_MODEL_ID environment variable is not set
-DEFAULT_MODEL_FALLBACK="meta-llama/llama-4-scout-17b-16e-instruct"
+DEFAULT_MODEL_FALLBACK="gpt-4o"
 # For codex script which uses a different default model
-DEFAULT_CODEX_MODEL_FALLBACK="meta-llama/llama-4-maverick-17b-128e-instruct"
+DEFAULT_CODEX_MODEL_FALLBACK="gpt-4o"
 
 if [ "$USE_CODEX" = "true" ]; then
     MODEL_ID=${DEFAULT_MODEL_ID:-"$DEFAULT_CODEX_MODEL_FALLBACK"}