MiMo-V2.6-Pro: benchmark results
Xiaomi's MIT-licensed omnimodal flagship MoE (1.02T total, 42B active) with a 1M-token context, served on the MiMo API and released as open weights the same day. Provider: Xiaomi. Released 2026-09-21. Access: Open.
Aggregate Rating 1823 ± 8, rank #49 of 1636 rated models, from 366 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SuperCLUE-Terminal - Overall | 64.65 | Score | 100 |
| VehicleArena - Basic Request Fulfillment | 90 | Score (0-100) | 100 |
| AA GDP.pdf - HR - Criterion Pass Rate | 90.13 | Criterion Pass Rate (%) | 99.6 |
| AA Long Context Reasoning | 86.33 | Accuracy (%) | 99.6 |
| AutomationBench-AA - Marketing - Raw Objective Completion | 91.11 | Raw Objective Completion (%) | 98 |
| Chatbot Arena (Text - Software & IT Services) | 1528 | Arena Score | 97.6 |
| Chatbot Arena (Text - Coding) | 1539 | Arena Score | 97.5 |
| Chatbot Arena (Text - English) | 1496 | Arena Score | 97.5 |
| Chatbot Arena (Text - Hard Prompts (English)) | 1515 | Arena Score | 97.2 |
| Chatbot Arena (Text - Expert) | 1531 | Arena Score | 96.7 |
| Epoch AI - Scicode | 60.88 | Score | 96.7 |
| AA SciCode | 60.88 | Accuracy (%) | 96.5 |
Interactive version: theaggregate.ai/model?slug=mimo-v2-6-pro · How It Works · Data refreshed daily, snapshot 2026-11-06.