Technology
Meta launches Muse Wonder models, its first attempt at frontier AI
Meta says its largest model beats Claude Fable 5.1 on some benchmarks, while training details remain undisclosed.
The short version
- Meta released Muse Wonder and Muse Wonder Full, which it describes as its first frontier-class models.[TechCrunch · Wired]
- Meta says the largest version outperforms Anthropic's Claude Fable 5.1 and ranks just below Claude Opus 5.5 on an aggregated index.[Wired]
- Muse Wonder Full is limited to Muse Code subscribers, while Muse Wonder costs $4.00 per million input tokens through Meta's API.[TechCrunch]
- Meta has not disclosed how the Muse Wonder models were built or trained.[Wired]
Key facts
- Meta released Muse Wonder and Muse Wonder Full AI models.[TechCrunch · Wired]
- Meta says the largest Muse Wonder version outperforms Anthropic's Claude Fable 5.1 and ranks just below Claude Opus 5.5 on the Artificial Analysis Intelligence Index.[Wired]
- Meta says Muse Wonder includes Active Memory and Active Focus features designed to keep agents aligned with stated goals.[Wired]
- Muse Wonder costs $4.00 per million input tokens and $20.00 per million output tokens through the Meta Model API.[TechCrunch]
- Muse Wonder Full is available only to Muse Code subscribers and is not served through the API.[TechCrunch]
- Meta did not disclose how the Muse Wonder models were built or trained.[Wired]
- Zuckerberg said in a post on X last month that Meta had delayed its Muse model releases to work on safety.[Wired]
What remains uncertain
- Meta's performance and benchmark claims are company-reported and not independently verified in the provided reporting.[TechCrunch · Wired]
- How the Muse Wonder models were built or trained remains undisclosed by Meta.[Wired]
Sources
Outlet counts describe coverage, not independent confirmation. Reports may share a wire service or original source.