Muse Spark 1.1 is the second release from Meta Superintelligence Labs, following the original Muse Spark that debuted in April 2026, and it ships with a multimodal reasoning design oriented toward agentic behavior. Coverage describes gains in tool use, computer use, coding, and multimodal understanding, paired with a 1M-token context window that allows long-horizon tasks to stay in a single session. Meta is positioning the model as a frontier-tier option alongside flagship releases from Anthropic, OpenAI, and Google, while framing its own benchmark numbers as surpassing prior products on Meta's internal evaluations. The release is also accompanied by the public preview of the Meta Model API, signaling Meta's entry into the paid developer API market alongside other major AI labs.

For practitioners, the practical appeal centers on agent workflows that mix reasoning, tool calls, structured output, and code generation, with the large context window supporting long documents, video captioning, and multi-step planning. The Decoder notes that the reported output price undercuts Grok 4.5, OpenAI, and Anthropic, making the model attractive for cost-sensitive deployments that still want frontier-class agent capability. Independent tooling such as OpenClaw has already added Muse Spark 1.1 as a Meta Model API endpoint, which broadens its reach into open-source agent platforms. As a closed-weights system accessed via API rather than local weights, it fits teams that prefer managed inference and are comfortable routing through the Meta Model API preview.