Introducing Muse Code and Muse Spark 1.2

Meta introduced Muse Code and Muse Spark version one point two, a major update to its coding-focused AI model. Per Simon Willison's Weblog, the release significantly improves code generation, debugging, codebase understanding, and end-to-end developer workflows. The model can now tackle whole-repository generation and large end-to-end projects, trained extensively on long-horizon coding tasks. Willison sees this as evidence of a broader trend in AI: long-sequence agentic tool calling—the ability to reason through complex tasks by repeatedly calling external tools—has become the most important capability in frontier models.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton