Document and chart understanding on CharXiv RQ
82.1Pass@1GPT-5.2 High
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5.2 High2026.06 | 82.1 | |
| Gemini-3-Pro High2026.06 | 81.4 | |
| Seed2.0 Pro2026.06 | 80.5 | |
| Seed2.0 Lite2026.06 | 79.9 | |
| Seed1.82026.06 | 71.4 | |
| Seed2.0 Mini2026.06 | 70.8 | |
| Gemini 1.5 Prothinking=true, decoding=greedy2025.05 | 69.9 | |
| Claude 3.7 Sonnetthinking=true, decoding=sampling2025.05 | 68.9 | |
| Claude-Opus-4.52026.06 | 65.5 | |
| Seed 1.5-VLthinking=true, decoding=greedy2025.05 | 60.2 | |
| Seed 1.5-VLthinking=false, decoding=greedy2025.05 | 59.8 | |
| OpenAI o1thinking=true, decoding=greedy2025.05 | 55.1 | |
| GPT-4othinking=false, decoding=greedy2025.05 | 52 | |
| Qwen 2.5-VL 72Bthinking=false, decoding=greedy2025.05 | 49.7 |