Meta Opens Muse Spark 1.1 to Developers, Targeting AI Coding Market

Meta is launching Muse Spark 1.1 via a new Model API, positioning its upgraded in-house AI model to compete directly in the developer tools and AI coding space. The new release promises stronger agentic workflows, bug detection, and native multimodal support.

After debuting its first in-house model in April, Meta is now making Muse Spark 1.1 available to developers through the new Meta Model API — a direct play for the competitive AI coding assistant market.

What's New in Muse Spark 1.1

Meta describes the upgrade as a "step-change" over the original Muse Spark, shaped by direct developer feedback. Key improvements include:

Advanced coding capabilities, including detection and automated fixing of complex bugs
End-to-end agentic workflow support across a broad range of applications, including multi-agent systems
Native multimodal perception spanning images, videos, and documents

Developer Access via the Model API

The Meta Model API is designed to let developers integrate Muse Spark 1.1 directly into AI coding tools and software pipelines. This positions Meta to compete with incumbents like OpenAI, Anthropic, and Google, all of which already have established API ecosystems targeting developers.

The move signals Meta's intent to shift from being primarily an open-source model provider — through its Llama family — to also competing in the hosted, commercial API layer where developer monetization is more direct.

Why It Matters

Meta's re-entry into proprietary AI model development is still recent, but the pace of iteration is notable. Going from an initial launch to a significantly upgraded API-accessible version within months reflects the pressure all major AI players are under to ship fast and capture developer mindshare.

With agentic AI workflows becoming a core battleground for enterprise adoption, Muse Spark 1.1's multi-agent support could be a meaningful differentiator — if it holds up in real-world developer testing.