Meta introduced Muse Ember on November 9 for the frequent, smaller steps within agent workflows: tool calls, routing, extraction and coding subtasks. The company lists it on the Meta Model API and in Muse Code. It is a separate model, not a renamed Wonder or Wonder Full.

Token prices depend on context

The exact muse-ember-1.0 API documentation charges $0.10 input, $0.004 cached input and $0.50 output per million tokens for requests with up to 200,000 input tokens, including exactly that boundary. Above it, the entire request uses $0.30/$0.012/$1.50 rates. Cached tokens and stored conversation history count toward the threshold.

The API card specifies a combined context of 1,048,576 tokens and up to 524,288 output tokens, with text, image, audio, video and PDF inputs. Prompt and output capacity are not independent unlimited allowances. Standard is the only Model API mode for Ember; there is no Contributor variant.

Efficiency claims need matched settings

Meta presents comparisons with Haiku and Luna across reasoning settings, alongside its own agent evaluations. Cost-per-task figures depend on model configuration, tokens and the stated pricing. Our comparison table keeps exact independent measurements separate from vendor results and does not invent an editorial score.