claude-flow 📚 New Wiki: Using Claude Code with OpenAI Open Models (GPT-OSS) and Qwen3-Coder

New Documentation Available

We've added comprehensive documentation for using Claude Code with alternative models including OpenAI's GPT-OSS models and Qwen3-Coder.

📖 Wiki Page

Using Claude Code with Open Models

What's Covered

Three Deployment Paths

Path A - Self-hosting on Hugging Face
- Deploy GPT-OSS-20B/120B or Qwen3-Coder on HF Inference Endpoints
- Configure Text Generation Inference (TGI) with OpenAI compatibility
- Direct integration with Claude Code via environment variables
Path B - OpenRouter Proxy
- Zero-DevOps solution
- Access to hundreds of models including GPT-OSS and Qwen
- Simple API key and URL configuration
Path C - LiteLLM Proxy
- Mixed fleet management
- Hot-swapping between Anthropic, GPT-OSS, Qwen, and Azure models
- Cost tracking and routing policies

Key Features Documented

Configuration

Minimal environment variable setup (ANTHROPIC_BASE_URL, ANTHROPIC_AUTH_TOKEN)
Model selection and routing
Streaming support via TGI's OpenAI-compatible API

Integration with Claude Flow

Using swarm orchestration with open models
Cost optimization strategies
Performance tracking across different models
Memory persistence regardless of model backend

Troubleshooting

Common issues and solutions
Performance optimization tips
Scaling considerations

Why This Matters

This documentation enables:

Cost Reduction: Use open-source models for appropriate tasks
Flexibility: Switch between models based on requirements
Control: Self-host sensitive workloads
Innovation: Experiment with cutting-edge open models

Technical Details

Environment Variables

export ANTHROPIC_BASE_URL="https://<your-endpoint>"
export ANTHROPIC_AUTH_TOKEN="<your-token>"
export ANTHROPIC_MODEL="gpt-oss-20b"

Claude Flow Enhancement

# Add MCP server
claude mcp add claude-flow npx claude-flow@alpha mcp start

# Use with open models
npx claude-flow@alpha swarm init --topology mesh --max-agents 5

References

The documentation includes links to:

Hugging Face Inference Endpoints documentation
OpenRouter API documentation
LiteLLM configuration guides
OpenAI's GPT-OSS announcement
Qwen3-Coder model cards

Related Issues

#596 - Claude Code hooks template variable fix
Part of alpha.87 release improvements

Labels

documentation, wiki, integration, open-models, gpt-oss, qwen, tutorial

Action Items

[ ] Review the wiki documentation
[ ] Test with your preferred deployment method
[ ] Provide feedback on any missing scenarios
[ ] Share your experience with different models

Wiki Link: https://github.com/ruvnet/claude-flow/wiki/Using-Claude-Code-with-Open-Models

This documentation is part of our effort to make Claude Flow more flexible and accessible with various AI models while maintaining the familiar Claude Code developer experience.