Muse Ember
Muse Ember is the smallest and most efficient model from Meta that’s designed for complex reasoning execution and the work agents repeat all day: tool calls, coding agent subtasks, routing,...
Specifications
| Lab | Meta |
|---|---|
| Context window | 1M tokens |
| Input price | $0.10/1M |
| Output price | $0.50/1M |
| Release | Nov 2026 |
Strengths
- Meta's smallest, most efficient reasoning model
- Ultra-budget pricing ($0.10 in / $0.50 out per M)
- 1M-token context; built for tool calls and agent subtasks
- Broad multimodal input (text, image, video, file, audio)
Weaknesses
- Small-model ceiling on hard reasoning vs flagship siblings
- Thin independent benchmark record at launch
Best for
- High-volume agent tool calls
- Coding-agent subtasks and routing
- Cost-sensitive production inference
In Depth: Muse Ember
Summary
Muse Ember is an AI model from Meta.
Released in Nov 2026. It takes text, image, video, file, and audio input and produces text output, with a context window of 1M tokens. Input pricing is $0.10 per 1M tokens and output is $0.50 per 1M tokens on OpenRouter.