Meta Ships Muse Spark 1.1 and Opens the Meta Model API
Meta Superintelligence Labs released Muse Spark 1.1, a multimodal reasoning model with a 1 million token context, and opened the Meta Model API in public preview, the first official developer API for Meta's own models.
Meta has models you can download and, until now, no official way to just call them. That changed on 9 July 2026, when Meta Superintelligence Labs released Muse Spark 1.1 and opened the Meta Model API in public preview. For the first time, Meta offers a first-party developer API for its own models, not only open weights you host yourself.
What happened
Meta Superintelligence Labs announced Muse Spark 1.1, a multimodal reasoning model aimed at agentic tasks, with a 1 million token context window. It ships in a “Thinking” mode inside the Meta AI app and on meta.ai, and, more importantly for builders, through the new Meta Model API, which entered public preview alongside it.
The API is the headline. Meta’s models have been available as open weights for years, which suited teams that could self-host but excluded everyone who wanted a hosted endpoint. A first-party Meta API means you can now reach Meta’s models the same way you reach Anthropic’s or OpenAI’s: a key, an endpoint, pay per use.
Why it matters for builders
This narrows a real gap. Until now, “use a Meta model” meant “run infrastructure,” which put Llama and its successors out of reach for teams that just wanted an API call. A hosted Meta endpoint changes the build-versus-buy calculus : you can prototype against Meta models without a GPU, then decide later whether to self-host for control or cost.
Two caveats keep this grounded. It is a public preview, so treat availability, pricing, and rate limits as provisional until general availability. And an official API is a different trade-off from open weights: you gain convenience and lose the data-residency and offline guarantees that made Meta’s open releases attractive in the first place. Choose the access path that matches your constraints, not the newest one.
Sources
- Meta, “Introducing Muse Spark and the Meta Model API”: https://ai.meta.com/blog/introducing-muse-spark-meta-model-api/
Further reading
- The 2026 LLM landscape : how the model families and their access models compare.
- Apple’s on-device Foundation Models : a different answer to the hosted-versus-local question.
- What is a mixture of experts? : the architecture behind most current multimodal models.