Overview
- Muse Ember, which Meta announced Monday, is available now on the Meta Model API and Muse Code with published pricing of $0.10 per million input tokens and $0.50 per million output tokens.
- The model is built for repeatable agent tasks such as tool calls, coding subtasks, data extraction, routing and long‑document parsing and it accepts text, images, audio and video inputs.
- Meta says Ember supports a 1 million token context window and up to 512,000 output tokens so agents can process entire codebases or large document sets in a single call.
- In Meta’s published Artificial Analysis comparisons Ember matches or exceeds Anthropic’s Haiku and OpenAI’s Luna on cost‑per‑task metrics, but independent third‑party benchmarks and broad enterprise validation remain limited.
- Meta trained Ember using Muse Wonder Full, says it passed internal safety checks, and is distributing the model through cloud partners such as Google Cloud and Oracle as part of a Muse stack that pairs larger planners with Ember for execution.