Multimodal Reasoning on ZeroBench main
11Pass@1Seed1.8
Evaluation results
| Method | Links | |
|---|---|---|
| Seed1.82026.03 | 11 | |
| Gemini 3-Pro2026.03 | 10 | |
| GPT-5.1 High2026.03 | 6 | |
| Claude Sonnet-4.52026.03 | 4 | |
| Gemini 1.5 Prothinking=true, decoding=greedy2025.05 | 3 | |
| Claude 3.7 Sonnetthinking=true, decoding=sampling2025.05 | 3 | |
| Gemini 2.5-Prosourced from technical report=true2026.03 | 3 | |
| Seed 1.5-VLthinking=true, decoding=greedy2025.05 | 2 | |
| Seed1.5-VL Thinking2026.03 | 2 | |
| Seed 1.5-VLthinking=false, decoding=greedy2025.05 | 0 | |
| OpenAI o1thinking=true, decoding=greedy2025.05 | 0 | |
| GPT-4othinking=false, decoding=greedy2025.05 | 0 | |
| Qwen 2.5-VL 72Bthinking=false, decoding=greedy2025.05 | 0 |