Meta Opens Muse Spark 1.1 to Developers with a Public API

Meta Superintelligence Labs releases Muse Spark 1.1, a multimodal reasoning model with million-token context, multi-agent orchestration, and a new developer API — positioning it directly against OpenAI, Anthropic, and Google.

Meta's Superintelligence Labs has released Muse Spark 1.1, a major upgrade to its in-house AI model that debuted in April. The new model brings significant advances in agentic reasoning, coding, and multimodal understanding, alongside a public API that puts Meta squarely in competition with the dominant AI labs.

Muse Spark 1.1 can orchestrate multi-agent systems — it acts as a main agent that gathers context, builds a plan, and delegates execution across parallel subagents. Each subagent understands its tools, adheres to its role, and knows when to escalate back to the main agent. The model manages a one-million-token context window, actively compacting information to retain critical steps across long workflows.

On the coding front, Meta says the model can diagnose and fix complex bugs in large codebases, implement features in enterprise-grade systems, and execute large code migrations. It also excels at computer-use tasks — navigating unfamiliar interfaces, writing automation scripts when faster, and clicking directly when simpler, all while adapting to information that changes mid-task.

The model is available now in "Thinking" mode through the Meta AI app and website. Alongside the release, Meta launched the Meta Model API in public preview for US developers, with $20 in free credits for new accounts. Early partners include Replit, Cline, and Box, with Cline's CEO noting the model's "strong tool use at a price point that makes it viable to run real coding workloads at scale."

This release follows this week's more controversial launch of Muse Image, which can pull other users' Instagram content into AI-generated photos. Together, the two launches signal Meta's accelerating pace in the AI race — and its willingness to ship products that push boundaries in both capability and public comfort.

Sources: Meta AI Blog, The Verge