Meta has flipped the switch on public developer access to its Muse Spark 1.1 AI model in the United States, moving decisively into the paid enterprise arena and squaring up to OpenAI and Anthropic. The shift signals a strategy pivot from open-source goodwill to commercial scale, with immediate implications for developers and brands.
Mark Zuckerberg marked the rollout by returning to X after three years, using the platform to amplify what Meta is framing as its biggest AI developer push of the year. The timing underscores how quickly the company wants traction for its next wave of AI experiences.
What the new model targets
Meta is pitching Muse Spark 1.1 as its strongest system yet for agentic and coding tasks. The company says it is designed to handle complex, multi-step workflows with minimal human oversight, closing the gap between conversational AI and practical software operations.
According to Meta, the upgraded model can write and debug code, use software and external tools, and understand text, images and videos. It is also rolling out in Thinking mode across the Meta AI app and website, setting the stage for richer, reasoning-driven interactions.
Meta says the same model will power upcoming AI features on WhatsApp, Instagram, Facebook and the companys smart glasses. That integration could turn everyday chats and feeds into front doors for automated planning, assistance and content creation.
Controlled distribution, wider access
Muse Spark debuted in April as a private preview for select partners. Meta has now expanded access, offering a public preview of the Muse Spark API through the Meta Model API for developers in the United States. Registration is open via the companys developer portal.
A Meta spokesperson said early partners already have access, while new users can join a waitlist and will be onboarded gradually. For now, the company is not distributing the model through third-party marketplaces such as OpenRouter.
This is going to be served on top of the computer infrastructure that weve built, said Alexandr Wang, positioning the decision as a bid for reliability and performance tied tightly to Meta’s stack.
Rapid release tempo
The API launch is the second major Muse move this week and follows Muse Image, the first image-generation model from Meta Superintelligence Labs. Together, they show Meta racing to assemble an integrated toolkit for developers, creators and consumers.
Muse Spark began as the inaugural text and reasoning model from Superintelligence Labs, formed last year to close the gap with leading AI rivals. The broader vision, described by Meta as delivering personal superintelligence, is now being operationalised across apps and APIs.
Competitive context and user impact
Opening developer access to Muse Spark 1.1 shifts Meta into a direct commercial contest with Anthropic and OpenAI. The company is moving beyond free releases toward enterprise-grade delivery, a step that could reshape how AI services are built and monetised on Meta’s platforms.
Speaking to Axios, Wang said the long-term goal is to make Meta AI more agentic, capable of independently handling tasks like planning a party or organising a vacation. He added that Meta’s biggest advantage is its billions of users and its deep understanding of their preferences and behaviour.
What developers can build now
For teams entering the public preview, Meta points to immediate opportunities tied to the models capabilities:
– Build coding assistants for real software projects
– Orchestrate external tools and software workflows
– Create multimodal agents that parse text, images and videos
– Automate multi-step tasks with minimal human input
The near-term question is execution speed. With access opening in the United States and Thinking mode arriving across Meta AI, the company is laying tracks for agentic experiences inside WhatsApp, Instagram, Facebook and smart glasses. The race now turns on who can convert those capabilities into everyday utility at scale.
