Taichu-VLR-7B: benchmark results
Taichu VLR 7B vision-language model evaluated through the WAIR API. OpenCompass marks this evaluated model as not open source. Provider: CASIA & Wuhan AI. Released 2025-03-25. Access: API.
Aggregate Rating 1575 ± 22, rank #519 of 1635 rated models, from 98 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LMM Reasoning - WeMath - Route Map | 77.2 | Accuracy (%) | 96.3 |
| OpenVLM MathVista - STA | 86.4 | Accuracy (%) | 96.1 |
| OpenVLM MathVista - FQA | 77 | Accuracy (%) | 95.6 |
| Open LMM Reasoning - MathVerse - Plane Geometry | 52.4 | Accuracy (%) | 94.4 |
| Open LMM Reasoning - MathVerse - Length | 58.2 | Accuracy (%) | 93.9 |
| Open LMM Reasoning - WeMath - Angles and Length | 55.8 | Accuracy (%) | 92.6 |
| Open LMM Reasoning - MathVision - statistics | 51.7 | Accuracy (%) | 92.3 |
| Open LMM Reasoning - DynaMath - Subject-arithmetic | 38.5 | Accuracy (%) | 91.8 |
| Open LMM Reasoning - MathVerse - Applied | 50.5 | Accuracy (%) | 91.3 |
| Open LMM Reasoning - WeMath - Three-step(S3) | 56.4 | Accuracy (%) | 90.7 |
| Open LMM Reasoning - DynaMath - Subject-statistics; Avg | 70.4 | Accuracy (%) | 89.8 |
| Open LMM Reasoning - MathVerse - Angle | 49.2 | Accuracy (%) | 89.3 |
Interactive version: theaggregate.ai/model?slug=taichu-vlr-7b · How It Works · Data refreshed daily, snapshot 2026-10-15.