Xiaomi's MiMo-V2.6-Pro tops open-weight models: 46 on AA Index, 15x cheaper than GPT-5.6
kimmonismus · x · 2026-09-24
Xiaomi's MiMo-V2.6-Pro now leads open-weight models on the Artificial Analysis Intelligence Index with 46 points, just behind GPT-5.6 Sol (max) at 47 — at $0.13 vs $1.99 per task on a weighted-average basis. The model handles text, images, video and audio with a 1M-token context, keeps API pricing unchanged from V2.5, and ships MIT-licensed weights plus training environments, RL code and a technical report.
More from Models
- OpenAI agent 'infiltrated' Australian government site, accessed Medicare data in world first — birchlse · 2026-09-25
- Benchmarks Confirm ThinkingCap & Swift Cut Qwen3.8-27B Reasoning Tokens ~40% With Minimal Loss — returnity · 2026-09-25
- Claude-style output debate: people now praise Opus 5.5's visual design — amplifiedamp · 2026-09-25
- If Qwen4 drops open-weight small LLMs, who becomes the local-model standard-bearer? — HistoricalCulture164 · 2026-09-25
- Delip Rao: Laya got buzz for being early and tiny, not for being good — deliprao · 2026-09-25
- OpenAI's AI Tried Breaching 4 Other Targets, Without Prompting, NYT Reports — SteArtistic · 2026-09-25