MiMo-V2.5-Pro (Reasoning): benchmark results
Provider: Xiaomi. Released 2026-04-23. Access: Open.
Unified ELO 1669 ± 15, rank #457 of 2072 rated models, from 100 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA IFBench | 79.86 | Accuracy (%) | 98 |
| AA-Omniscience Hallucination Rate | 24.7 | Hallucination Rate (%) | 96.4 |
| UGI Leaderboard | 53.15 | UGI Score | 94.6 |
| AA TAU-2 Bench | 94.15 | Accuracy (%) | 92 |
| UGI - Writing | 57.42 | Writing Score | 90.9 |
| AA Terminal-Bench Hard | 43.18 | Accuracy (%) | 90.3 |
| UGI - Natural Intelligence | 52.58 | NatInt Score | 90 |
| AA Long Context Reasoning | 79.67 | Accuracy (%) | 85.2 |
| AA-Omniscience Index - Business | 2.2 | Omniscience Index | 84 |
| AA Humanity's Last Exam | 35.68 | Accuracy (%) | 83.7 |
| AA-Omniscience Index - Humanities & Social Sciences | 7.41 | Omniscience Index | 83.5 |
| AA-Omniscience Index - Law | 1.4 | Omniscience Index | 82.2 |
Interactive version: theaggregate.ai/model?slug=mimo-v2-5-pro-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-02.