WeThink-Qwen2.5VL-7B: benchmark results
Provider: WeChat AI. Access: Open.
Aggregate Rating 1579 ± 27, rank #1257 of 10916 rated models, from 150 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM AI2D - Parts of the Earth | 88.5 | Accuracy (%) | 98.4 |
| OpenVLM MMMU - Electronics | 86.7 | Accuracy (%) | 98.2 |
| OpenVLM MMMU - History | 83.3 | Accuracy (%) | 96.5 |
| OpenVLM MathVista - STA | 85.7 | Accuracy (%) | 94.9 |
| OpenVLM MMBench V1.1 CN - OCR | 98.9 | Accuracy (%) | 94 |
| OpenVLM OCRBench - Doc-oriented VQA | 89.5 | Correct Answers (%) | 93.8 |
| OpenVLM MMBench V1.1 EN - Celebrity Recognition | 92.5 | Accuracy (%) | 93.5 |
| OpenVLM MMBench V1.1 CN - Object Localization | 86.7 | Accuracy (%) | 92.9 |
| OpenVLM MMMU - Geography | 66.7 | Accuracy (%) | 92.8 |
| OpenVLM MathVista - FQA | 75.5 | Accuracy (%) | 92.5 |
| OpenVLM MMBench V1.1 EN - FP-S | 91.2 | Accuracy (%) | 91.7 |
| OpenVLM MM-Vet - Recognition | 73.3 | Score | 91.2 |
Interactive version: theaggregate.ai/model?slug=wethink-qwen2-5vl-7b · How It Works · Data refreshed daily, snapshot 2026-11-08.