Meta Introduces Muse Ember for Repetitive Agent Work

Meta says Muse Ember (muse-ember-1.0) is designed for repeated agent tasks such as tool calls, coding subtasks, routing, extraction and long-document parsing. Meta reports strong results on several agent and coding benchmarks, while its own pricing and availability details describe access through Muse Code and the Meta Model API.

What happened

In its announcement, Meta introduced Muse Ember as its smallest and most efficient model for complex reasoning execution. The company says it was trained using Muse Wonder Full and is intended to handle frequent, structured steps in agent workflows.

What changed

Meta lists Muse Ember as available as muse-ember-1.0 through Muse Code and the Meta Model API, with support for the Responses API. Its stated rates are $0.10 per million input tokens, $0.004 per million cached input tokens and $0.50 per million output tokens. Meta gives a 1M-token context window and a maximum output of 512K tokens.

Meta’s benchmark table reports Muse Ember (max) scores of 37.8 on AutomationBench-AA and 40.9 on Terminal-Bench 4.0. It also reports 27.2 on FrontierCode 2.0-AA, below Claude Haiku 5.5’s 34.9 in that comparison. Meta says its cost-per-task estimates use proposed pricing; these results and comparisons are company-reported, not independent findings in this article.

Creator impact

For visual creators, the announcement is relevant mainly to production systems that use agents to organize assets, route requests, extract information from briefs or handle repetitive coding and document work. Meta says Muse Ember can understand text, images, audio and video, and can work with screenshots and scanned documents. That could matter when a workflow needs a model to interpret mixed media, but the announcement does not establish image-generation or video-generation features.

Limitations

The source does not provide independent testing of Muse Ember, and benchmark outcomes may not predict performance in a creator’s own workflow. Meta’s claimed pricing and availability should be checked against its current API terms. The article also does not establish access in every region or detail operational requirements.

An honest AiPix use case

AiPix is an AI Agent-powered image and video creation platform built around AI Canvas. Muse Ember may be relevant as a model for agent-side tasks such as parsing a creative brief or routing repetitive work, but the source does not say it is integrated with AiPix. It is not evidence of a new AiPix feature; creators can continue using AiPix’s existing creation paths for image and video work.