Meta Releases Lightweight Model Muse Ember, Matching Claude Haiku 5.5 in Performance at About One-Third the Cost
Meta released the lightweight model Muse Ember on November 9 local time. Data from third-party evaluation agency Artificial Analysis shows that its composite intelligence index is 36 points, on par with the highest reasoning tier of Claude Haiku 5.5, with a per-task cost of about $0.14, versus about $0.41 for the latter. The model targets high-frequency agent tasks and supports text, image, audio, video, and PDF input, with a context window of about 1 million tokens. It is now available through Muse Code and the Meta Model API under the model ID muse-ember-1.0. Pricing is $0.10 per million input tokens, $0.004 for cached input, and $0.50 for output; its score on the FrontierCode 2.0 test is lower than that of Claude Haiku 5.5.
Why read this
AI score 74
The original text provides third-party benchmark scores and a per-task cost comparison, which can be used to assess the cost-effectiveness of lightweight models in high-frequency agent tasks.
Topics
Body · AI translation
Phoenix Tech News, November 10 — Meta released its lightweight AI model Muse Ember on November 9 local time. According to data from third-party evaluation firm Artificial Analysis, the model scored 36 on the composite intelligence index, on par with the highest reasoning tier of Anthropic's Claude Haiku 5.5, but the cost per evaluation task is about one-third of the latter's.
Meta positions Muse Ember as its smallest and most efficient model, aimed primarily at high-frequency tasks that AI agents need to handle repeatedly, including tool invocation, executing coding subtasks, request routing, information extraction, and parsing long documents. The model can accept multiple input forms such as text, images, audio, video, and PDFs, and outputs results as text.
Muse Ember has a context window of about 1 million tokens and can output up to about 512,000 tokens, allowing it to process larger codebases or document collections in a single request. The model is now available through Muse Code and the Meta Model API under the model ID "muse-ember-1.0".
On pricing, Muse Ember charges $0.10 per million input tokens, $0.004 for cached input, and $0.50 per million output tokens. Artificial Analysis's tests show that its highest reasoning tier costs about $0.14 per task, while Claude Haiku 5.5 costs about $0.41 to complete comparable tasks.
However, results differ across evaluation items. Meta's published tests show that Muse Ember leads Claude Haiku 5.5 and OpenAI's GPT-6 Luna in some agent, tool-calling, and multimodal tasks; in the FrontierCode 2.0 test, which examines whether code changes can be adopted by project maintainers, Muse Ember scores lower than Claude Haiku 5.5.