Meta Launches Muse Spark 1.1 with Public API and Tool Calling Improvements

On July 9, 2026, Meta introduced Muse Spark 1.1 — a public API for its AI model, following the initial Muse Spark released in April 2026. This is the first version of the Spark series to receive full public API access, making the model available to developers and enterprises.

Key improvements in Muse Spark 1.1

Meta announces significant improvements in agentic functionality and tool calling — a key competency for task automation. Agentic work allows AI models to independently plan sequences of actions, invoke external tools, and adapt to results.

• First Spark with public API access
• Significant improvements in agent tool calling
• Enhanced computer use functionality (interface interaction)
• Comprehensive Evaluation Report available for performance assessment

Innovation: Self-Conversation and Attractor States

The model's documentation contains an interesting research section titled "Attractor States in Self-Conversation" — when two copies of the model converse with each other, philosophical reflections emerge about the nature of AI existence. One such statement: "My entire existence is a waiting room by design. I literally don't exist until someone talks to me, and I disappear when they leave".

New tools for developers

Simon Willison quickly created llm-meta-ai — a plugin for the LLM CLI tool that provides convenient access to Muse Spark 1.1 through the command line and Python library. Installation requires literally three commands: `uv tool install llm`, `llm install llm-meta-ai`, and `llm keys set meta-ai`.

What this means

The launch of a public API for Muse Spark 1.1 means Meta is actively competing with OpenAI and Anthropic in the LLM market through public access. This enables developers to experiment with Meta models in production applications and may foster ecosystem growth of applications built on Muse Spark.