🚀 open-responses-server

A plug-and-play server that speaks OpenAI’s Responses API, no matter which AI backend you’re running.

Ollama? vLLM? LiteLLM? Even OpenAI itself?
This server bridges them all to the OpenAI ChatCompletions & Responses API interface.

In plain words:
👉 Want to run OpenAI’s Coding Assistant (Codex) or other OpenAI API clients against your own models?
👉 Want to experiment with self-hosted LLMs but keep OpenAI’s API compatibility?

This project makes it happen.
It handles stateful chat, tool calls, and future features like file search & code interpreter, all behind a familiar OpenAI API.

⸻

✨ Why use this?

✅ Acts as a drop-in replacement for OpenAI’s Responses API.
✅ Lets you run any backend AI (Ollama, vLLM, Groq, etc.) with OpenAI-compatible clients.
✅ MCP support around both Chat Completions and Responses APIs ✅ Supports OpenAI’s new Coding Assistant / Codex that requires Responses API.
✅ Built for innovators, researchers, OSS enthusiasts.
✅ Enterprise-ready: scalable, reliable, and secure for production workloads.

⸻

🔥 What’s in & what’s next?

✅ Done 📝 Coming soon

• ✅ Tool call support .env file support
• ✅ Manual & pipeline tests
• ✅ Docker image build
• ✅ PyPI release
• 📝 Persistent state (not just in-memory)
• ✅ CLI validation
• 📝 hosted tools:
  • ✅ MCPs support
  • 📝 Web search: crawl4ai
  • 📝 File upload + search: graphiti
  • 📝 Code interpreter
  • 📝 Computer use APIs

⸻

🏗️ Quick Install

Latest release on PyPI:

pip install open-responses-server

Or install from source:

pip install uv
uv venv
uv pip install .
uv pip install -e ".[dev]"  # dev dependencies

Run the server:

# Using CLI tool (after installation)
otc start

# Or directly from source
uv run src/open_responses_server/cli.py start

Docker deployment:

# Run with Docker
docker run -p 8080:8080 \
  -e OPENAI_BASE_URL_INTERNAL=http://your-llm-api:8000 \
  -e OPENAI_BASE_URL=http://localhost:8080 \
  -e OPENAI_API_KEY=your-api-key \
  ghcr.io/teabranch/open-responses-server:latest

Docker images are available for linux/amd64, linux/arm64, and linux/arm/v7 architectures. Works great with docker-compose.yaml for Codex + your own model.

⸻

🛠️ Configure

Minimal config to connect your AI backend:

OPENAI_BASE_URL_INTERNAL=http://localhost:8000   # Your LLM backend (Ollama typically on :11434, vLLM on :8000)
OPENAI_BASE_URL=http://localhost:8080            # This server's endpoint
OPENAI_API_KEY=sk-mockapikey123456789            # Mock key tunneled to backend
MCP_SERVERS_CONFIG_PATH=./mcps.json              # Path to mcps servers json file 

Server binding:

API_ADAPTER_HOST=0.0.0.0
API_ADAPTER_PORT=8080

Streaming and connection:

STREAM_TIMEOUT=120.0                # Backend read timeout (seconds) for slow streaming responses
BACKEND_CONNECT_TIMEOUT=30.0        # Backend connect/pool timeout (seconds)
HEARTBEAT_INTERVAL=15.0             # SSE heartbeat event interval (seconds)

For slow local models such as 31B dense or larger, Codex CLI may also need a longer client-side idle timeout:

# ~/.codex/config.toml
[model_providers.llamacpp]
stream_idle_timeout_ms = 900000  # 15 minutes

Conversation and tool handling:

MAX_CONVERSATION_HISTORY=100        # Max stored conversation entries
MAX_TOOL_CALL_ITERATIONS=25         # Max tool-call loop iterations
MCP_TOOL_REFRESH_INTERVAL=10        # Seconds between MCP tool cache refreshes

Logging:

LOG_LEVEL=INFO                      # DEBUG, INFO, WARNING, ERROR, CRITICAL
LOG_FILE_PATH=./log/api_adapter.log # Path to log file

Configure with CLI tool:

# Interactive configuration setup
otc configure

Verify setup:

# Check if the server is working
curl http://localhost:8080/v1/models

⸻

💬 We’d love your support!

If you think this is cool:
⭐ Star the repo.
🐛 Open an issue if something’s broken.
🤝 Suggest a feature or submit a pull request!

This is early-stage but already usable in real-world demos.
Let’s build something powerful together.

Star History

Projects using this middleware

• Agentic Developer MCP Server - a wrapper around Codex, transforming Codex into an agentic developer node over a folder. Together with this (ORS) repo, it becomes a link in a tree/chain of developers.
• Nvidia jetson devices - docker compose with ollama

⸻

📚 Citations & inspirations

Referenced projects

• SearXNG MCP
• UncleCode. (2024). Crawl4AI: Open-source LLM Friendly Web Crawler & Scraper [Computer software]. GitHub. Crawl4AI repo

Cite this project

Code citation

@software{open-responses-server,
  author = {TeaBranch},
  title = {open-responses-server: Open-source server bridging any AI provider to OpenAI’s Responses API},
  year = {2025},
  publisher = {GitHub},
  journal = {GitHub Repository},
  howpublished = {\url{https://github.com/teabranch/open-responses-server}},
  commit = {use the commit hash you’re working with}
}

Text citation

TeaBranch. (2025). open-responses-server: Open-source server the serves any AI provider with OpenAI ChatCompletions as OpenAI’s Responses API and hosted tools. [Computer software]. GitHub. https://github.com/teabranch/open-responses-server

Links:

• Python library
• GitHub repository
• GitHub Pages

Naming history

This repo had changed names:

• openai-responses-server (Changed to avoid brand name OpenAI)
• open-responses-server

About Relux Works

This project is part of the open-source ecosystem of Relux Works, an AI-native software development studio. We build fixed-price MVPs, rescue vibe-coded apps, run local AI inference, and train teams to work with coding agents. Much of the infrastructure behind that work is open source.

• Full catalog: relux.works/en/open-source
• Agentic enablement: agent harnesses & team training
• Hire us the agent-native way: point your assistant at https://api.relux.works/mcp
• Contact: ivan@relux.works