MiMo-V2.6-Pro
Xiaomi · 1T total / 42B active (MoE)
Open-weight omni-modal with long context
A 1T-A42B omni-modal model released 21 September 2026, with open weights, 1M context and text/image/video/audio input. Video 94 is the joint-best open-weight video score here; perception 94 and agents 92 are also top tier, for a composite of 91.4 that replaces the previous edition’s MiMo-V2.5. Documents 92 and ecosystem 85 remain behind the Qwen family for enterprise delivery.
Last updated Oct 21, 2026Editorial scoring
Bottom line
Eight-dimension breakdown
editorial mapping · 0–100- MiMo-V2.6-Pro91.4
Weights: perception 20 / documents 18 / charts 14 / video 14 / agents 14 / generation 6 / Chinese 6 / ecosystem 8. Definitions and sources on the methodology page.
Public benchmarks
as reported, not converted| Benchmark / metric | Value | Note |
|---|---|---|
| 参数规模 | 1T-A42B | Official spec |
| 模态 | 4 | Text/image/video/audio |
| 上下文 | 1M | Official spec |
Public scores come from different versions and prompts and are not interchangeable with our composite; they are one evidence input.
Scenario mapping
five stress tests, one scale| Scenario | Score | Observation |
|---|---|---|
| Chinese receipt OCR | 91 · Stable on clean receipts | |
| Chart data extraction | 89 · Reliable on standard charts | |
| 30-min video QA | 94 · Top band among open weights | |
| Mobile GUI control | 90 · Well tuned to own devices | |
| Chinese cultural context | 89 · Strong on everyday scenes; average academic context |
Strengths and limitations
Strengths
- Video 94 is the joint-best open-weight score, and 1M context keeps long-video QA from over-chunking.
- Four input modalities at 1T-A42B: rare omni-modal capacity among open weights.
- ¥3/1M output tokens with free weights keeps both edge and cloud costs down.
Limitations
- The Xiaomi community licence is not OSI open source; commercial terms need legal review.
- Documents 92 and ecosystem 85 trail the Qwen family on complex layouts and enterprise delivery.
- Grade B: public benchmarks are mostly vendor-reported, with little third-party reproduction.
Price and real cost
List price
¥3/M tokens
≈ ¥3 / 1M output tokens; weights free. For 1,000 images at 1,024px, ~1,100 image tokens each and 300 output tokens, the monthly cost is about ¥1.8 (assuming input at 25% of output price).
Use your volumeAbout MiMo-V2.6-Pro
Is MiMo-V2.6-Pro open source?
Licence: Open weights (Xiaomi community licence). Weights are available, but check the exact terms and filing requirements before commercial use.
How much does MiMo-V2.6-Pro cost?
≈ ¥3 / 1M output tokens; weights free. It ranks #11 with a composite of 91.4. Real bills depend on resolution, tiling, caching and volume discounts.
What is MiMo-V2.6-Pro best for?
Long-video understanding and content tagging; Products spanning edge and cloud; Self-hosted omni-modal pipelines。
Why does MiMo-V2.6-Pro score 91.4?
The composite weights eight dimensions: perception 20%, documents 18%, charts 14%, video 14%, agents 14%, generation 6%, Chinese 6%, ecosystem 8%. Its strongest dimension is Visual perception & grounding; its weakest is Ecosystem, cost & availability.
Personal ranking · Not an official score
Every ranking, score and comparison on this page is an independent personal ranking, not an official score, and represents no vendor, institution or regulator. Scores are a subjective mapping of public evidence under our own weights — change the weights and the order changes. Disagree, or spotted a factual error? Contact the site owner: jizhi2024@foxmail.com — we check each report and log corrections in the changelog.