Xiaomi open-sources MiMo-V2.6: MoE models with 1M context, MIT license
AdinaYakup · x · 2026-09-22
Xiaomi's @XiaomiMiMo released two new Sparse MoE models, both with 1M context, native omni (text/image/video/audio) and MIT license:
- MiMo-V2.6 Pro RL (max capability): 1.02T total / 42B active params, with strong coding and agent scores — 71.9 DeepSWE, 53.1 AutomationBench, 89.9 on Terminal Bench 2.1
- MiMo-V2.6 Flash RL (max efficiency): 309B total / only 15B active params, staying close to Pro on many agent benchmarks
The poster claims one may be the best open model yet; the collection is live on Hugging Face.
More from coding & agent
- YC-backed Roma turns your to-do list into an agent command center across Gmail and Notion — testingcatalog · 2026-11-12
- Boubou: open-source CLI turns your laptop into a fully local voice AI hub — alexcovo_eth · 2026-11-12
- User automates entire video clipping workflow with Nous Research's Hermes agent — Teknium · 2026-11-12
- Tool-call failure rates jump 14x across providers, charts pinpoint inference stack updates — remilouf · 2026-11-12
- As models improve, writing good evals gets harder and more important — amankhan · 2026-11-12
- Prime agent ships with day-one multi-agent communication, runs in Minecraft unmodified — sarahcat21 · 2026-11-12