Skip to content
NewA calmer, faster ZeroHour
StoryModel releaseinfo

Meta releases hosted Muse Ember for high-volume agent work

5 sources · 5 articlesFirst reported Updated ()2
What's new

November 10: Simon Willison says short-prompt pricing matches GPT-6 Luna and Claude Haiku 5.5; longer-prompt pricing is less straightforward.

Merged summary · x-ai/grok-4.7 · rewritten as coverage arrivesSummaries by Vani AI

Meta’s private-weight Muse Ember is live at $0.10 and $0.50 per million tokens for routine agent tasks.

Meta Superintelligence Lab released Muse Ember, its smallest and most cost-efficient model, for tool calls, coding subtasks, routing, extraction, and long-document parsing. It is available only through the Meta Model API and Muse Code, with a 1 million-token context and up to 512,000 output tokens; weights stay private. List pricing is $0.10 per million input tokens and $0.50 per million output tokens. Simon Willison says that rate matches GPT-6 Luna and Claude Haiku 5.5 for shorter prompts, while longer-prompt pricing is less straightforward. Meta cites Artificial Analysis to claim more intelligence per dollar than Haiku and Luna; The Decoder puts cost near $0.14 per task, MarkTechPost near $0.136.

Key facts
  • Fourth Muse model, after Spark, open-source Glimmer, and Wonder.
  • Accepts text, images, audio, and, per Decoder and MarkTechPost, video.
  • Cached input is listed at $0.004 per million tokens.
  • Artificial Analysis score 36; AutomationBench-AA 37.8 at max reasoning.
  • FrontierCode 2.0-AA: Ember 27.2 percent, Haiku 34.9 percent.
  • Early testers include Box, Cline, Oracle, Cloudflare, and You.com.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

Timeline

dated events in the reporting · oldest first
  1. Nov 9, 2026Meta releases hosted Muse Ember model · TechCrunch

What each outlet adds

beyond the shared facts
  • Hacker NewsTerminal-Bench 4.0 at 40.9; some cost figures are estimates.
  • TechCrunchCost-competitive with DeepSeek V4.1 Flash and GLM 5.3 Flash.
  • The DecoderAbout $0.14 per task versus $2.39 for Muse Spark 1.3.
  • MarkTechPostCached input at $0.004; names Cline, Oracle, Cloudflare, You.com.
  • Simon WillisonShort-prompt price parity with Luna and Haiku; longer pricing unclear.

Coverage

oldest first · one row per article · UTC
  1. · 1d ago Hacker News · AIFirst report

    Meta releases Muse Ember for repeated agent and coding work

    It is on the Meta Model API at $0.10 per million input tokens and $0.50 per million output tokens.

  2. · 1d ago TechCrunch · AI

    Meta releases Muse Ember, a lightweight model for agent work

    Meta prices Ember at 10 cents per million input tokens and pairs it with Muse Wonder for volume.

  3. · 1d ago The Decoder

    Meta launches Muse Ember, a small model for agent routine tasks

    Meta prices it at $0.10 per million input tokens and says it scores 36 on Artificial Analysis for about $0.14 a task.

  4. · 19h ago MarkTechPost

    Meta releases Muse Ember, a hosted 1M-context agent model

    Weights stay private; Meta prices it at $0.10/$0.50 per million tokens and about $0.136 per AA task.

  5. · 18h ago Simon Willison

    Meta AI launches Muse Ember, its smallest efficient model

    Short prompts match GPT-6 Luna and Claude Haiku 5.5 at $0.10 and $0.50 per million tokens.