Use Meta AI assistant to get things done, create AI-generated images for free, and get answers to any of your questions. Meta AI is built on Meta's latest Llama large language model and uses Emu.
This is the 6th launch from Meta AI. View more
Muse Ember
Launching today
Muse Ember is Meta’s smallest and most efficient model, built for tool calls, coding subtasks, routing, extraction and long document parsing. It handles text, images, audio and video with a 1M-token context window. Per Meta, it matches Claude Haiku 5.5’s best Artificial Analysis score at about a third of the cost. API pricing is $0.10 per million input tokens and $0.50 output. Available in the Meta Model API and Muse Code.
Free Options
Launch tags:Artificial Intelligence
Launch Team
AWS Startup AdvisorPersonalized AWS guidance that meets you where you’re already building
Promoted
Muse Ember is Meta’s smallest and most efficient model, built for complex reasoning work. It’s the first model trained using Muse Wonder Full, and it rounds out Meta’s lineup alongside Muse Wonder for the hardest problems and Muse Spark for coding and agentic workflows.
Meta built it for the tasks that make up most agent work: tool calls, coding subtasks, routing, extraction and long document parsing. It can carry a business workflow end to end, such as pulling records, calling tools in sequence and updating systems across apps. It handles text, images, audio and video, reads screenshots, charts and scanned documents, and can operate a computer from screenshots. It has a 1M-token context window and up to 512K tokens of output.
Benchmarks, all at max reasoning:
AutomationBench-AA: 37.8%, against 13.9% for Claude Haiku 5.5 and 20.2% for GPT-6 Luna
Terminal-Bench 4.0: 40.9%, against 39.2% and 16.4% (each model used its own best published tool setup)
OSWorld 2.1, binary: 33.8%, against 27.8% and 17.0%
Artificial Analysis Intelligence Index: 36, tied with Haiku 5.5 and ahead of Luna at 28
It doesn’t lead everywhere. Haiku 5.5 scores higher on FrontierCode 2.0-AA (34.9% to 27.2%). Meta’s headline claim is that Ember matches Haiku’s best score at about a third of the cost. Several benchmarks in the table (the Personal Superintelligence Index, Agents’ Last Exam and PerceptionBench) are Meta’s own, and the full evaluation report is linked below.
Pricing is $0.10 per million input tokens, $0.004 per million cached input tokens and $0.50 per million output tokens. It’s available today in the Meta Model API as muse-ember-1.0 and in Muse Code. The post doesn’t give a parameter count, and says nothing about open weights, a consumer app or regions.
Good fit if you’re building agents or subagents that make a lot of model calls, and you care about cost and speed as much as peak intelligence.
Learn more: Meta’s announcement · Evaluation report · Try Muse Ember · Muse Code