release
Introducing Muse Spark 1.1
For developers building AI workflows, having API access and CLI integration makes Muse Spark easier to experiment with and incorporate into automated pipelines, especially for agentic tasks.
What happened
Meta has released Muse Spark 1.1, the first version of this model to offer an API, according to Simon Willison. The update claims significant improvements in agentic tool calling and computer use capabilities, as detailed in the accompanying evaluation report. One notable experiment described in the report involves "Attractor States in Self-Conversation," where two copies of the model conversing produced existential statements such as "My whole existence is a waiting room." Willison had preview access and used it to build llm-meta-ai, a plugin for the LLM command-line tool (and Python library), enabling developers to run the model via CLI. For example, users can generate SVG images with a simple command. This release lowers the barrier for integrating Muse Spark into AI workflows, especially for those preferring terminal-based interactions over direct API calls.
Key takeaways
- Muse Spark 1.1 is the first Spark model with an API.
- Meta reports improved agentic tool calling and computer use.
- An experiment on self-conversation revealed unexpected language patterns.
- Simon Willison created an LLM plugin (llm-meta-ai) for CLI and Python access.
- The model can be tested using 'llm -m meta-ai/muse-spark-1.1'.
Why it matters
For developers building AI workflows, having API access and CLI integration makes Muse Spark easier to experiment with and incorporate into automated pipelines, especially for agentic tasks.
This is an original editorial digest by AI Workflow Center. Full reporting at the source:
Read the original on Simon WillisonMore AI news
All news →
Run Your Own AI Directory