Kimi-VL-A3B: benchmark results
Provider: Moonshot. Released 2025-04-10.
Aggregate Rating 1499 ± 11, rank #2374 of 10916 rated models, from 385 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LMM Reasoning - WeMath - InsufficientKnowledge (Loose) | 99.6 | Accuracy (%) | 100 |
| Open LMM Reasoning - WeMath - RoteMemorization (Loose) | 100 | Accuracy (%) | 100 |
| Open LMM Reasoning - WeMath - RoteMemorization (Strict) | 100 | Accuracy (%) | 99.4 |
| Open LMM Reasoning - DynaMath - Subject-solid geometry; Avg | 55.3 | Accuracy (%) | 98 |
| Open LMM Reasoning - DynaMath - Subject-arithmetic; Avg | 52.7 | Accuracy (%) | 92.9 |
| OpenVLM MathVista - STA | 84.7 | Accuracy (%) | 92.3 |
| OpenVLM OCRBench - Doc-oriented VQA | 88.5 | Correct Answers (%) | 91.7 |
| Open LMM Reasoning - MathVerse - Volume | 39.2 | Accuracy (%) | 91.3 |
| UltraVR - Whole-Slide Imaging | 53.6 | Accuracy (%) on the whole-slide pathology imaging domain (mu | 90 |
| Open LMM Reasoning - DynaMath - Level-elementary school; Avg | 64.6 | Accuracy (%) | 87.8 |
| OpenVLM MM-Vet - Spatial Awareness | 75.1 | Score | 87.6 |
| OpenVLM MathVista - MWP | 80.1 | Accuracy (%) | 86.1 |
Interactive version: theaggregate.ai/model?slug=kimi-vl-a3b · How It Works · Data refreshed daily, snapshot 2026-11-08.