Moonshot's Kimi K 3.1 Pro surfaces: 2.8T-param MoE, weights on Hugging Face
thursdai_pod · x · 2026-11-06
According to the ThursdAI podcast account, Moonshot has dropped Kimi K 3.1 Pro: 2.8 trillion total parameters in a 104B-active MoE architecture, with weights already on Hugging Face. The poster teased notable benchmark results, though specifics weren't shown. Third-party report, unconfirmed by official channels.
More from Models
- Blogger Cites Epoch Odds: GPT-6 Astra 72% Likely to Beat Opus 5.5 — haider1 · 2026-11-06
- Ed Witten cites Claude Opus 5.5 as assistant in serious math paper — teortaxesTex · 2026-11-06
- Google Ultra subscriber blocked from Antigravity by region detection — lxfater · 2026-11-06
- Palisade's Jeff Ladish: skip the 40 benchmarks, ask if agents finish real work — FinanceYF5 · 2026-11-06
- AA Index 31: 594B total / 17B active params, Apache 2.0 open source — AIFlow_ML · 2026-11-06
- Tencent Hunyuan Hy4 hits limits on day one, WorkBuddy rushes 10x compute expansion — TencentHunyuan · 2026-11-06