Meta · 2026-05-01 · notable
Llama Stack v0.8.0 — Native Anthropic Messages API, Gemini Interactions, 91% OpenAI Compatibility
Open GenAI Stack adds /v1/messages passthrough for Anthropic APIs, Google Interactions API frontend for Gemini, raises OpenAI API conformance from 89.1% to 91.2%, and adds conversation compaction for long agent sessions.
Llama Stack now speaks Anthropic's Messages API and Google's Interactions API from a single unified GenAI stack.
Key specs
| GitHub stars | 8,362 |
|---|---|
| Contributors | 27 |
| Openai conformance | 0.912 |
What is it?
Llama Stack is Meta's open-source, unified deployment stack for Llama-family models and an expanding set of third-party models. Version 0.8.0 landed May 1, 2026, adding multi-provider API compatibility as a core focus.
How does it work?
The v0.8.0 release implements the /v1/messages endpoint in both passthrough mode (forwarding to Anthropic directly) and translation mode (converting Anthropic-format requests to Llama Stack internals). The Google Interactions API frontend provides a similar bridge for Gemini. The OpenAI API conformance suite now covers 91.2% of the spec. Conversation compaction automatically summarizes old context in long Responses API sessions to keep agents within context limits.
Why does it matter?
By natively supporting Anthropic, Gemini, and OpenAI API formats, Llama Stack is moving from a Llama-specific tool to a universal GenAI interface that lets teams switch providers without rewriting application code.
Who is it for?
Enterprise teams wanting a provider-agnostic deployment layer for LLMs, especially those running on-premise Llama models alongside cloud-hosted Claude or Gemini.
Try it
pip install llama-stack==0.8.0