Xiaomi's MiMo-V2.6-Pro lands on Fireworks: 1.02T MoE, 1M context, MIT license
sophiamyang · x · 2026-09-26
MiMo-V2.6-Pro-RL, Xiaomi's flagship agentic omnimodal model, is now available as on-demand deployment on Fireworks AI. The 1.02T-parameter sparse MoE (42B active) supports text/image/video/audio input with 1M-token context under an MIT license, and reportedly scores highest among open-weight models on the AA Intelligence Index (46). It scales a unified RL pipeline across coding, agents, vision and cybersecurity tasks. No fine-tuning, serverless or function calling supported.
More from Models
- GPT-6-Astra-Max Tops BALROG Game Benchmark at 68.3%, NetHack Depth Hits 13.2 — burny_tech · 2026-09-26
- Claude Opus proves surprisingly good at both making IK rigs and animating with them — andrew_n_carr · 2026-09-26
- Two years since o1-preview: reasoning tokens went from novelty to frontier standard — ArtificialAnlys · 2026-09-26
- Open-source JevBench adds $49/$99 paid priority benchmark runs with 48-hour turnaround — airesearch12 · 2026-09-26
- Ling Tiny 3.0 on a 2017 laptop: 8B model hits 10 tok/s, hints at edge AI future — netherreddit · 2026-09-26
- Watch Claude Opus 5.5 generate 100 matplotlib plots in one go — goodside · 2026-09-26