Meta has introduced Muse Spark 1.1 and a public-preview Meta Model API, positioning the updated model for coding, agentic workflows, and developer access. Axios and CNBC report that the release is part of Meta’s effort to compete more directly with AI coding and assistant products from OpenAI and Anthropic.
Meta introduced Muse Spark 1.1 and opened public-preview access to its Meta Model API, according to the company’s announcement and reports from Axios and CNBC.
The release gives developers access to Meta’s updated general-purpose AI model through an API, moving the company further into a market where OpenAI, Anthropic, Google and others are competing to provide models for software development, assistant-style workflows and multimodal tasks.
In its announcement, Meta described Muse Spark 1.1 as an updated model available through the Meta Model API. Axios reported that Meta emphasized improvements in coding and longer agentic tasks, while CNBC described Muse Spark 1.1 as Meta’s strongest model yet for agentic and coding work.
“Agentic” tasks generally refer to workflows where an AI system can carry out multi-step actions toward a user’s goal, such as writing and revising code, using tools, or handling extended instructions. The sources provided do not include independent benchmark verification, so Meta’s performance claims should be read as company claims unless and until outside testing confirms them.
The public-preview API is significant because it gives developers a more direct way to test Meta’s model outside the company’s consumer products. CNBC reported that the API is moving from private preview toward public-preview developer access, while Axios reported that the developer version was released alongside the Muse Spark update.
CNBC framed the launch as Meta’s entry into the AI coding market in an effort to chase Anthropic and OpenAI. Those companies have become prominent providers of models used by developers for code generation, debugging, and software-assistant workflows.
Meta’s announcement also positions the model as part of a broader push to provide AI infrastructure to developers, not just AI features inside Meta-owned apps. The company’s decision to expose Muse Spark through an API suggests it wants third-party developers to build on top of its models and compare them directly with rival commercial APIs.
Axios reported that the update highlights coding and longer agentic tasks, two areas where enterprise and developer demand has grown. CNBC likewise reported that Meta is emphasizing Muse Spark 1.1 as a model for agentic and coding work.
Because Muse Spark 1.1 is being offered in public preview, developers should expect the product to continue changing. The source material does not provide a full general-availability date, final pricing details, or independent comparisons against the newest models from competing providers.
That matters because model announcements often include internal benchmarks or selected comparisons, while developers and businesses usually need broader information: reliability, latency, tool-use behavior, pricing, data-handling terms, and performance on real coding tasks.
Meta’s announcement establishes the company’s direction: make Muse Spark 1.1 accessible to developers and compete in the market for coding and agentic AI systems. Axios and CNBC both report that the release is part of Meta’s broader attempt to strengthen its position against leading AI model providers.
For now, Muse Spark 1.1’s importance is less about a settled performance ranking and more about access. With the Meta Model API entering public preview, developers can begin testing whether Meta’s newest model is useful for coding, longer task execution, and AI application development.
Meta introduced Muse Spark 1.1 and opened public preview access to its Meta Model API, according to the company’s announcement and reports from Axios and CNBC.
Meta focuses on coding and agentic work In its announcement, Meta described Muse Spark 1.1 as an updated model available through the Meta Model API.
Axios reported that Meta emphasized improvements in coding and longer agentic tasks, while CNBC described Muse Spark 1.1 as Meta’s strongest model yet for agentic and coding work.
Continue reading
