Muse Spark 1.1 by Meta is a multimodal reasoning model for agentic tasks with zero-shot tool generalization, computer use, coding, and 1M-token context.
Capabilities, design details, and architectural traits
Muse Spark 1.1 is a multimodal reasoning model from Meta Superintelligence Labs, built specifically for agentic tasks rather than conversational use. It delivers major gains in tool and computer use, coding, and multimodal understanding over its predecessor, Muse Spark.
The model is available in Thinking mode in the Meta AI app and on meta.ai, with developer access through a public preview of the Meta Model API.
| Trait | Detail |
|---|---|
| Zero-shot tool generalization | Generalizes to new native tools, MCP servers, and custom skills without prior training on each interface |
| Planning and orchestration | Handles personal agentic tasks that require coordinating across a range of external apps and services |
| Computer use workflow | Decides when to write automation scripts versus interact with interfaces directly, generating batches of actions at each step |
| Active memory management | With a 1M-token context window, it compacts information and retains critical steps while discarding the rest |
| Coding with verification loop | Writes code, takes screenshots to check results, traces failures to the source, and fixes them in iteration |
| Adaptive execution | Notices when context changes mid-task and adjusts plans without user intervention |
Muse Spark 1.1 navigates unfamiliar desktop interfaces with minimal human intervention. It maintains context across extended sessions and adapts to evolving requirements. For coding, it works on real enterprise codebases - diagnosing bugs, implementing features, and executing large code migrations - rather than toy problems.
The model's training approach for computer use is distinctive: it writes scripts when automation is faster and clicks through interfaces when direct interaction is simpler, choosing the appropriate strategy per situation.
Independent evaluations · Artificial Analysis
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | Meta | Anthropic | Anthropic |
| Release Date | July 9, 2026 | July 24, 2026 | June 9, 2026 |
| Knowledge Cutoff | - | May 2026 | - |
| Context & Limits | |||
| Context Window | 1M | 1M | 1M |
| Pricing (per 1M tokens) | |||
| Input Pricing | $1.25 Best Input Pricing | $5 | $10 |
| Output Pricing | $4.25 Best Output Pricing | $25 | $50 |
| Modalities | |||
| Inputs | textimagevideo | textimage | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 53.2 | 63.1 Best Intelligence Index | 62.1 |
| Coding Index | 71.3 | 78.0 Best Coding Index | 76.5 |
| Agentic Index | 39.7 | 59.2 Best Agentic Index | 56.6 |
Graduate-level reasoning and expert Q&A evaluation.
Extremely difficult logical reasoning and knowledge.
Logical reasoning over long context windows.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
