Xiaomi open-sources MiMo-V2.6: 1.02T-param Pro brings near-frontier multimodal reasoning
teortaxesTex · x · 2026-09-22
The author jokes that GRPO is "God's Recommended Policy Optimization," still going strong after 2.5 years — if it doesn't work, you're not doing enough of it. The quoted tweet carries the news: Xiaomi's MiMo-V2.6 ships in Flash and Pro variants — Flash at 309B total / 15B active params, Pro at 1.02T total / 42B active — promising near-frontier multimodal reasoning, with weights for both now on Hugging Face.
More from Fun
- "Thanks to the entire reels industry for the data": video model training sources meme — xiaosun86 · 2026-11-12
- Endless screensaver of self-assembling quantum cryostats, made with Gemini 4 Carbon — hapticdata · 2026-11-12
- Dev Community Bakes Power-Outlet-Shaped Cookies to Celebrate a Birthday — threepointone · 2026-11-12
- Claude Code's Cryptic "Misting..." Status Sparks 25-Minute Jokes — _Stocko_ · 2026-11-12
- Viral "judge scolds woman" images are AI fakes: garbled nameplate and stenographer slip-ups — aakashgupta · 2026-11-12
- Claude writes a Bach-style organ fugue in LilyPond with a single prompt — IgorCarron · 2026-11-12