Rumor: open-weight Claude Sonnet 5.5-xhigh could land within 10 days
AaronBergman18 · x · 2026-10-25
Aaron Bergman speculates, based on a "No publicly available API" leaderboard, that Anthropic may release open weights for Sonnet 5.5-xhigh in 10 days, cautioning its edge is mostly on benchmarks. The chart shows Claude Opus 5.5 (max) with fallback leading at 58, with several Claude 5.5 variants, Muse Wonder, and Kimi K3.1 Pro (55) at the top; Grok 4.7 (xhigh) at 47, DeepSeek V4.1 107T at 46, GLM-5.3 at 45. Unverified rumor.
More from Models
- Claude notices it's Sunday for a European user and tells them to go touch grass — aiamblichus · 2026-10-25
- Claude Models Now GA on Google Cloud with New Anthropic-Managed Deployment Option — emmanuelvivier · 2026-10-25
- GPT-6 Astra Hits 39% on OpenDS-World Live Web Deep Search Benchmark vs 21.1% Human Average — emmanuelvivier · 2026-10-25
- Study: Grok picks Republicans 92%, GPT-6.1 Sol picks Democrats 97% — emmanuelvivier · 2026-10-25
- Microsoft's MAI-DxO solves 83.2% of complex cases vs physicians' 36.1% — emmanuelvivier · 2026-10-25
- Domain breakdown: Kimi K3.1 Pro matches Opus 5.5 High in most fields, lags in legal — AaronBergman18 · 2026-10-25