Document Understanding on FireRedBench (test)
0.8185Overall ScoreQwen3.5-397B-A17B
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Qwen3.5-397B-A17BModel Category=General VLMs2026.03 | 0.8185 | 0.205 | 0.9414 | 0.7193 | 0.7825 | 0.383 | |
| Gemini-3.0 ProModel Category=General VLMs2026.03 | 0.7968 | 0.169 | 0.8011 | 0.7582 | 0.8273 | 0.353 | |
| Qwen3-VL-235BModel Category=General VLMs2026.03 | 0.7904 | 0.208 | 0.9083 | 0.6709 | 0.7452 | 0.366 | |
| PaddleOCR-VL-1.5Model Category=Pipeline OCR Systems2026.03 | 0.7647 | 0.291 | 0.9237 | 0.6615 | 0.7439 | 0.453 | |
| FireRed-OCR-2BModel Category=End-to-End OCR Models2026.03 | 0.7462 | 0.248 | 0.8302 | 0.6563 | 0.723 | 0.43 | |
| GLM-OCRModel Category=Pipeline OCR Systems2026.03 | 0.7433 | 0.309 | 0.8253 | 0.7135 | 0.7993 | 0.456 | |
| dots.ocrModel Category=End-to-End OCR Models2026.03 | 0.7293 | 0.24 | 0.8253 | 0.6025 | 0.6408 | 0.419 | |
| GPT-5.2Model Category=General VLMs2026.03 | 0.6809 | 0.238 | 0.6633 | 0.6174 | 0.68 | 0.38 | |
| Qwen3-VL-2BModel Category=End-to-End OCR Models2026.03 | 0.6558 | 0.283 | 0.7519 | 0.4985 | 0.5566 | 0.388 | |
| DeepSeek-OCR 2Model Category=End-to-End OCR Models2026.03 | 0.6161 | 0.29 | 0.5878 | 0.5506 | 0.5942 | 0.437 |