Meta Releases Muse Spark 1.1 AI Model with 1 Million-Token Context Window to Target Enterprise Market

Meta Platforms (NASDAQ: META) announced Thursday the release of Muse Spark 1.1, a new multimodal reasoning model from its Meta Superintelligence Labs division. The model is designed for autonomous "agentic" workflows and features a massive one-million-token context window, alongside significant upgrades to its coding and computer control functions, signaling a major push into the enterprise AI market.

In a parallel move to attract developers, Meta also launched a public preview of its Meta Model API, granting external developers access to the Muse architecture for the first time. The company stated the model is now live for consumers in "Thinking" mode on its Meta AI application and website.

Key Capabilities and Enterprise Focus

According to Meta, the defining feature of Muse Spark 1.1 is its ability to operate within multi-agent networks. The model can function as both a "primary agent" and a "subagent," allowing it to decompose large tasks and delegate them to parallel subagents to improve processing speed and efficiency.

Meta is positioning the model as a direct competitor to other frontier engineering models. The company reported that internal testing on its Meta Internal Coding Bench showed Muse Spark 1.1 outperforms its predecessor and is competitive against unnamed industry alternatives. Its capabilities reportedly include autonomously diagnosing bugs, implementing new features into large enterprise codebases, and executing complex code migrations.

Developer Adoption and Partner Feedback

The announcement highlighted early adoption from several API partners, indicating initial traction in the developer community. Key launch partners include:

Ad

• Replit: CEO Amjad Masad stated, "The combination of a million-token context window, deep multimodal support, and advanced coding capabilities fundamentally changes how we think about scale."
• Cline: CEO Saoud Rizwan noted the model's tool-use flexibility and competitive pricing for high-volume developer workloads.
• Box: Yashodha Bhavnani, VP of AI Products, said Spark 1.1 performed "head-to-head with top-tier frontier models" on Box’s internal enterprise evaluation benchmarks.

Beyond coding, Muse Spark 1.1 can automate workflows across desktop applications by writing its own scripts or interacting directly with user interfaces, a feature Meta calls "computer use."

Safety Audits and Broader Rollout

Addressing industry safety concerns, Meta confirmed the model was audited under its Advanced AI Scaling Framework. The evaluations tested for catastrophic risks, including chemical and biological threats, cybersecurity vulnerabilities, and potential "loss of control" scenarios. Meta stated the model operates safely within its established margins and that a full evaluation report will be published soon.

The launch of Muse Spark 1.1 coincides with the general release of Muse Image, Meta’s latest standalone image-generation model.