Meta releases compact Muse Ember model

On November 9, 2026, Meta released Muse Ember (muse-ember-1.0) — the smallest and most affordable multimodal model in the Muse lineup, featuring a 1 million token context window and designed for high-volume agentic tasks.

Compiled by Sergey KostenchukPublished 2026-11-10Updated 2026-11-10

2026-11-10 Research Anthropic

🤖 Meta releases compact Muse Ember model

On November 9, 2026, Meta released Muse Ember (muse-ember-1.0) — the smallest and most affordable model in the Muse lineup for agentic routine tasks: tool calls, coding subtasks, routing, and data extraction. The context window is 1 million tokens, with output up to 524,288; multimodality includes text, images, audio, and video. Pricing is $0.10/$0.50 per 1 million input/output tokens.

🌍 According to Artificial Analysis, Ember scores 35.8 points at ~$0.136 per task; Claude Haiku 5.5 scores 36.2 at ~$0.411. Intelligence is slightly lower, but the cost is three times cheaper: API calls, parsing, and code subtasks in agentic pipelines can be offloaded to the cheaper model.

👤 muse-ember-1.0 is available via the Meta Model API and Muse Code, as well as in the Google Cloud Model Garden (Preview). It is a cost-effective option for routine tasks: screenshots, recordings, and long documents. Evaluations: Meta's report and one measurement by Artificial Analysis — no independent verifications are available.

Source 1: https://research.meta.ai/blog/muse-ember

Sources