Meta / Muse Spark 1.1 Released July 9, 2026

Quick answer

Muse Spark 1.1 is an agent model with a preview API.

Meta positions Muse Spark 1.1 for agent orchestration, computer use, coding, and multimodal work. It is available in Meta AI, while developer access is explicitly a Meta Model API public preview.

01

Agent orchestration

Plan, gather context, delegate to subagents, and escalate from a subagent role when needed.

02

Computer use

Choose between scripts, direct interface actions, and batched operations across changing workflows.

03

Coding

Work across complex codebases, multi-turn harnesses, planning modes, and validation steps.

04

Multimodal reasoning

Inspect visual and audio inputs and retain useful details across a long agentic workflow.

Context behavior

One million tokens is not one million remembered facts

Meta says the model actively manages a one-million-token window, retrieves earlier actions, and compacts context. Evaluate what survives compaction and whether tools receive the right state on your real workflow.

Developer boundary

Public preview is a lifecycle, not a production guarantee

Use current Meta developer sources for endpoint shape, pricing, limits, data policy, compatibility, and availability. Do not infer any of them from a launch-level capability summary.

Primary source

Introducing Muse Spark 1.1 Meta's July 9, 2026 model and Model API public-preview record.
Muse Image Released image generation, editing, and agentic media tools. Muse Video Coming-soon media preview; not a released API surface. Muse family guide Compare lifecycle, task, and access boundaries. GPT-5.6 Compare another current reasoning and agentic model family.

Muse Spark 1.1 questions, answered

Muse Spark 1.1 is Meta's multimodal reasoning model for agentic tasks, introduced on July 9, 2026. Meta highlights tool and computer use, coding, multimodal understanding, and multi-agent orchestration.
Meta says Muse Spark 1.1 is available in Thinking mode in the Meta AI app and on meta.ai. Developer access launched through the new Meta Model API in public preview.
No. Meta describes the Model API as a public preview. Treat its endpoints, limits, pricing, compatibility, and production commitments as current-source checks rather than durable facts from this page.
Meta's launch article says Muse Spark 1.1 can actively manage a one-million-token context window, retrieve earlier information, and compact context to preserve critical steps.
Meta says the model can generalize to tools, MCP servers, and custom skills, and can operate as either a coordinating main agent or a scoped subagent. Real integrations still need permission, security, and reliability testing.
This page does not claim a Flowith model mapping. Meta AI and Meta Model API availability do not establish Flowith access; check the live Flowith workspace separately.