Leak: Meta drops Muse Wonder models claiming SOTA agents, coding, $4/$20 pricing
ns123abc · x · 2026-10-13
A leak claims Meta's Superintelligence Lab has released Muse Wonder and Muse Wonder Full, billed as its most capable models:
- SOTA on personal agents, memory, and coding benchmarks: beats Fable 5.1 on DeepSWE, Opus 5.5 on long-horizon coding, and GPT-6 Astra on most evals
- Trained on a gigawatt cluster named Prometheus
- Native 1M-token context window at inference
- Priced at $4 / $20 per 1M tokens
Note: from a third-party account, unverified pending official confirmation; a model card link is attached.
More from Models
- Muse Wonder accused of benchmaxxing: costs $18 vs free SWE 2 and performs worse — lyraxana · 2026-10-13
- Writer ditches $200/mo Opus 5.5 for Gemini 4 over one reason: it writes well — iamtrask · 2026-10-13
- Alleged Gemini 4 Argon Builds 3D Chess in One Shot, User Stunned — cgarciae88 · 2026-10-13
- Meta's Muse Wonder model goes live on OpenRouter — jeff_weinstein · 2026-10-13
- Muse senior engineer benchmark results pulled over flawed run, retest underway — danshipper · 2026-10-13
- "Zero chance" a Meta model competes with Google, OpenAI or Anthropic right now — iruletheworldmo · 2026-10-13