MiMo-VL-7B-RL-2508: benchmark results
Provider: Xiaomi. Access: Open.
Aggregate Rating 1620 ± 16, rank #833 of 10916 rated models, from 68 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HyperTrack | 76.41 | Step Exact Match (%; predicted action type and parameters bo | 96 |
| VURB - General Video Understanding | 57.1 | Majority-vote pairwise accuracy (%; eight judgments with the | 85.7 |
| MMMU Benchmark | 70.6 | Validation Score | 80.3 |
| RobotEQ-Video - Real Videos | 76.1 | Real-subset Avg (%; action-properness judgment: given an ego | 78.7 |
| SciVQR | 69.2 | Overall accuracy (%) with zero-shot chain-of-thought prompti | 78.6 |
| SciVQR - Chemistry | 74.7 | Chemistry accuracy (%) with zero-shot chain-of-thought promp | 78.6 |
| SciVQR - Geography | 70.6 | Geography accuracy (%) with zero-shot chain-of-thought promp | 78.6 |
| SciVQR - Mathematics | 65.5 | Mathematics accuracy (%) with zero-shot chain-of-thought pro | 78.6 |
| SciVQR - Physics | 67.5 | Physics accuracy (%) with zero-shot chain-of-thought prompti | 78.6 |
| VURB - Long Video Understanding | 62 | Majority-vote pairwise accuracy (%; eight judgments with the | 78.6 |
| FREAK - Attribute | 48.46 | Accuracy (%) on the attribute questions of FREAK's 1,799 que | 76.5 |
| EXPLORE-Bench - Long Sequences | 45.17 | Unified score (0-100): 0.25 x object coverage + 0.35 x attri | 74.1 |
Interactive version: theaggregate.ai/model?slug=mimo-vl-7b-rl-2508 · How It Works · Data refreshed daily, snapshot 2026-11-08.