---
id: "meta/muse-ember"
also: ["meta/muse-ember-20261109", "vercel/meta/muse-ember"]
card_usd_per_m: "0.1 in, 0.5 out, at meta"
cheapest_list_usd_per_m: "0.1 in, 0.5 out, at google-vertex/global on openrouter"
cheapest_effective_usd_per_m: "0.0267 in, 0.4999 out, at google-vertex/global on openrouter"
context: 1048576
max_output: 943718
input: ["text", "image", "video", "file", "audio"]
output: ["text"]
released: "2026-11-09"
reasoning: ["max", "xhigh", "high", "medium", "low", "minimal"]
api: ["/api/v1/chat/completions", "/api/v1/completions", "/api/v1/messages", "/api/v1/responses"]
params: ["include_reasoning", "max_tokens", "reasoning", "reasoning_effort", "repetition_penalty", "response_format", "seed", "stop", "structured_outputs", "temperature", "tool_choice", "tools", "top_k", "top_p"]
endpoints: 3
gateways: "openrouter (2), vercel (1)"
no_training: 3
trains: 0
retains_prompts: 1
---
# Meta: Muse Ember

Muse Ember is Meta’s multimodal language model for conversational, reasoning, and agentic applications. It supports text and image understanding, tool calling, structured outputs, and document input, with a 1,048,576-token context window for working with large prompts and extended conversations. Implicit prompt caching supports reuse of shared input across requests.