Megadose Models, research, and what moved.

Our Smallest, Most Efficient Model: Muse Ember

· HN · Frontpage AI ·
Meta prices Muse Ember at $0.10 per million input tokens and $0.50 per million output tokens for repeatable agent work.

The model is available through Muse Code and the Meta Model API, targeting tool calls, coding subtasks, extraction and document parsing. It supports a 1M-token context window, up to 512K output tokens, and text, images, audio and video. Meta’s reported results put it near Claude Haiku 5.5’s best Artificial Analysis score at roughly one-third the cost per task, though Haiku leads on the FrontierCode engineering benchmark. Some evaluations are internal, and the cost comparison uses proposed pricing.

HN · Frontpage AI's note

score 6

Categories: Model Releases

Discussions

  • hn · 250 points · 105 comments
  • hn · 256 points · 107 comments
  • hn · 259 points · 108 comments
  • hn · 263 points · 110 comments