Muse Ember on BenchGecko · Meta's smallest Muse model at $0.10 / $0.50
Meta released Muse Ember on 2026-11-09. BenchGecko lists it at $0.10 input and $0.50 output per M tokens with a 1M token context and an Artificial Analysis Index of 36.
What launched
Meta Superintelligence Labs released Muse Ember on November 9, 2026 as the smallest model in its Muse family, next to Muse Spark. Meta positions it for the high volume work agents repeat: tool calls, coding subtasks, routing, extraction and long document parsing. It is available on the Meta Model API, Google Cloud and Oracle Cloud. Weights are not released.
Specs
- Context · 1M tokens
- Max output · 512K tokens
- Inputs · text, images, audio and video
Pricing on BenchGecko
For prompts up to 200K tokens, Muse Ember lists at $0.10 input, $0.50 output and $0.004 cached input per million tokens. Above 200K the rates rise to $0.30, $1.50 and $0.012. Meta estimates about $0.13 of model cost per Artificial Analysis Intelligence Index task at max effort.
Vendor reported benchmarks
Meta reports these figures at max reasoning effort. BenchGecko records vendor claims as published.
- Artificial Analysis Intelligence Index · 36 (Claude Haiku 5.5 36 · GPT-6 Luna 28)
- AutomationBench-AA · 37.8 (Haiku 5.5 13.9 · Luna 20.2)
- Agents' Last Exam · 61.6 (Haiku 5.5 57.0 · Luna 57.1)
- Terminal-Bench 4.0 · 40.9 (Haiku 5.5 39.2 · Luna 16.4)
- Coding Agent Index-AA · 46 (Haiku 5.5 37 · Luna 31)
- FrontierCode 2.0-AA · 27.2 (Haiku 5.5 34.9 · Luna 25.2)
- OSWorld 2.1 · 69.5 partial · 33.8 binary
Sources
Source: BenchGecko
Written automatically from BenchGecko's tracked data for this period; every number comes from the changelog, model, mindshare and pricing datasets. Checked against the source data before publishing.