MiMo-V2.5 — benchmark results
Xiaomi's open multimodal MiMo-V2.5 MoE (310B total, 15B active) with a 1M-token context. Provider: Xiaomi. Released 2026-04-23. Access: Open.
Unified ELO 1724 ± 15, rank #230 of 1776 rated models, from 107 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| PinchBench | 91.87 | Success Rate (%) | 91.2 |
| AA Terminal-Bench Hard | 41.67 | Accuracy (%) | 88.4 |
| CritPt | 3.7 | Accuracy (self-reported) | 87.6 |
| Design Arena (Website) | 1288 | Elo | 87.5 |
| Design Arena (Data Viz) | 1279 | Elo | 87 |
| Artificial Analysis Intelligence Index | 37.2 | Intelligence Index | 86.5 |
| Design Arena (Game Dev) | 1291 | Elo | 86.4 |
| Design Arena (UI Components) | 1294 | Elo | 86.4 |
| AA TAU-2 Bench | 90.64 | Accuracy (%) | 85.4 |
| AA Humanity's Last Exam | 25.16 | Accuracy (%) | 84.8 |
| AA GPQA Diamond | 84.95 | Accuracy (%) | 83.9 |
| DystopiaBench | 34.5 | Dystopian Compliance Score (0-100) | 83.7 |
Interactive version: theaggregate.ai/model?slug=mimo-v2-5 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.