v0.1.34

🚀 Release Highlights — v0.1.34 (2026-01-05)

🌐 OpenAI-Compatible Server

• Chat Completions API: Run MassGen as a local server that provides a Chat Completions-compatible API
• OpenAI SDK Compatible: Any OpenAI SDK client can connect seamlessly
• Custom Configs: Use default config or specify your own with --config

🔍 Dynamic Model Discovery

• Authenticated API Calls: Groq and Together backends fetch available models via API instead of hardcoded lists
• OpenAI-Compatible Endpoints: Supports standard model discovery for dynamic backend configuration

🖥️ WebUI Improvements

• File Diffs: View workspace file changes with diff highlighting
• Answer Refresh: Polling-based updates for real-time answer display
• Performance: Faster workspace browser with optimized caching

🤖 Subagent System Reliability

• Status Tracking: Improved subagent status monitoring and error handling
• Cancellation Recovery: Better handling of cancelled subagent operations
• Context Handling: Fixed context and media handling for subagent workflows

---

📖 Getting Started

• Quick Start Guide: Try the new features today
• Try These Examples:

  # Start OpenAI-compatible server with default config
  massgen serve --host 0.0.0.0 --port 4000
  
  # Or specify a custom config
  massgen serve --config @examples/basic/multi/three_agents_default
  
  # Use with any OpenAI SDK client
  curl http://localhost:4000/v1/chat/completions \
    -H "Content-Type: application/json" \
    -d '{"model": "massgen", "messages": [{"role": "user", "content": "Explain multi-agent systems in LLMs"}]}'

What's Changed

• docs: update slides post v0.1.32 by @sonichi in #719
• fix: Fix responses compression input issue by @ncrispino in #746
• fix: Modified the code in base_with_custom_tool_and_mcp.py to make logfire truly optional. by @randombet in #738
• docs: clarify backend model discovery behavior (MAS-163) by @shadowstar654 in #669
• feat: OpenAI Compatible Chat Completions server by @maxim-saplin in #689
• fix: fix workspace browser artifact viewer by @ncrispino in #741
• feat: Fix subagent responses by @ncrispino in #740
• docs: Docs for v0.1.34 by @Henry-811 in #751
• feat: v0.1.34 by @Henry-811 in #742

New Contributors

• @shadowstar654 made their first contribution in #669

Full Changelog: v0.1.33...v0.1.34