Meta launches Muse Wonder models: SOTA agents, Active Memory, gigawatt training
ren_hongyu · x · 2026-10-13
Meta AI researcher Ren Hongyu announced Muse Wonder and Muse Wonder Full, a frontier model series claiming SOTA on internal personal and professional agentic tasks, framed as a step toward personal superintelligence.
Key points:
- Combines Active Memory and Active Focus to track what matters to users and work toward their goals
- Trained at the gigawatt-scale Prometheus data center; caps a 14-month ground-up rebuild of Meta's AI stack following the predictable-scaling approach set by Muse Spark
- Availability: Muse Wonder is live via the Meta Model API and Muse Code; Muse Wonder Full for Muse Code subscribers. It will later power Muse experiences, artifact generation, and specialized distilled models
Benchmark charts show Muse Wonder Full (max) at 59.1 on the Personal Superintelligence Index vs Claude Opus 5.5 (55.2) and GPT-6 Astra (54.5), leading DeepSWE 1.1 (79.0) and SWE-Marathon (57.5), though trailing Opus 5.5 on ProgramBench (59.0 vs 70.5).
Related event: Meta Unveils Muse Wonder, Its Most Capable Frontier Models Yet(45 posts)→
More from coding & agent
- Meta launches Muse Wonder models, tops SWE-Marathon at 57.5% — acbuller · 2026-10-13
- Meta's superintelligence lab launches Muse Wonder, a frontier model tuned for personal agents — alexandr_wang · 2026-10-13
- Muse Code CLI goes live with one-line curl install for Muse Wonder — alexandr_wang · 2026-10-13
- Perplexity: agent grep exploration hits 0.627 answer accuracy vs ~0.37 for semantic search — perplexity_ai · 2026-10-13
- Perplexity swaps long-lived sandboxes for ephemeral ones on S3-backed storage — perplexity_ai · 2026-10-13
- How Perplexity serializes concurrent agent submits with head checks and replay — perplexity_ai · 2026-10-13