Muse Spark 1.1
Language modelPrevious generationCard summarizedMuse Spark is one of the most price-efficient, high-intelligence models available for developers to build with.
Active (original version; 1.3 recommended for new work)● Re-checked 12 Oct 2026 · 58 facts · 1 correction · 1 unconfirmedSource: Muse Spark 1.1 Evaluation Report (112 pp)
Use it via API
Where to call it
- Meta Model API
muse-spark-1.1
Supports
- Image input
- PDF input
- Structured outputs
- Reasoning control
- Tool calling
- Web search
- Prompt caching
Good for
- Personal agentic tasks: planning and orchestration across external apps, MCP servers and custom skills
- Multi-agent work: as main agent delegating to parallel subagents, or as a subagent
- Computer use across multiple applications, mixing scripts and direct interface actions
Overview
Muse Spark 1.1
Muse Spark is one of the most price-efficient, high-intelligence models available for developers to build with.
- Cost
- $1.25 in
- Context
- 1M
- Input / output
- $1.25 / $4.25
- Max output
- —
- Knows up to
- —
- Released
- 8 Jul 2026
Muse Spark 1.1 updates Muse Spark 1.0, the model behind Meta AI, and is the first Muse model offered to developers, through the Meta Model API. Before mitigations Meta finds it at the 'high' risk threshold for chemical & biological risk and cannot rule out 'high' for cybersecurity; with its deployed safeguards Meta rates residual risk 'moderate or lower' in every domain. Meta centres the assessment on the API because it exposes the broadest agentic affordances. p.1, p.5–9
muse-spark-1.1- Input / output per 1M
- $1.25 / $4.25
- Context window
- 1M
- Max output
- —
- Knows up to
- —
- Reasoning
- Always on; reasoning_effort minimal to xhigh
- Weights
- Closed
About this card. A model card is the maker's own account of a model: what it's for, how it was trained and tested, and where it falls short. This page condenses Meta Superintelligence Labs's Muse Spark 1.1 Evaluation Report. Every claim carries its source page.