MiMo-V2.6-Pro
Xiaomi-reported: DeepSWE v1.1 71.9, Terminal-Bench 2.1 89.9, AutomationBench 53.1, CyberGym 94.0. MOPD (distilled) variant added ~Sept 27.
- Context window
- 1,000,000 tokens
- Max output
- 128,000 tokens
- Input
- text, image, video, audio
- Output
- text
- License
- mit
- Pricing
- input: $0.435 · cached input: $0.0036 · output: $0.87 (per 1M tokens (USD; list price ¥3 / ¥6)) source
- Verified
- 2026-09-29
How to call it
| Provider | Model id | Endpoint / URL | Docs |
|---|---|---|---|
| Xiaomi MiMo API | mimo-v2.6-pro | — | docs |
| Hugging Face | — | huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL | — |
| OpenRouter (Pro-UltraSpeed tier, ~20x output speed, $4.35 / $8.70) | see https://openrouter.ai (Xiaomi MiMo V2.6 Pro UltraSpeed) | — | — |
Notable capabilities (2)
- Top open-weights model on the Artificial Analysis Intelligence Index: Scored 46 at launch (Sept 2026), tying Grok 4.7 and ahead of DeepSeek V4.1 Flash (39). source
- Omnimodal 1T-parameter MIT-licensed MoE: 1.02T total / 42B active, 70 layers (60 SWA + 10 global), 1M context; text, image, video and audio input under MIT. source
Timeline entry
- Xiaomi releases MiMo-V2.6 Pro (1.02T MoE) and Flash under MIT license; Pro becomes the top open-weights model on Artificial Analysis ★★★★
On Sept 21–22, 2026 Xiaomi released the MiMo-V2.6 series under the MIT license: Pro (1.02T total / 42B active MoE), Flash (~311B / 15B active), a Pro-UltraSpeed serving tier and a 9B Qwen distill. Both main models have 1M context and omnimodal input. Pro scored 46 on the Artificial Analysis…
Other Xiaomi models
MiMo-V2.6-Flash · Xiaomi-Robotics-1 (XR-1, 5B) · Xiaomi-Robotics-U0 (38B) / U0-4B · Xiaomi-Robotics-0 (4.7B VLA)