Multimodal Reasoning on RBench-V
32.4Overall ScoreCanvas-of-thought
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Canvas-of-thoughtModel Name=OpenAI GPT-52026.02 | 32.4 | 1 | 48.9 | 1.1 | 47.1 | 1.4 | 27.2 | 0.9 | 16.7 | 0.9 | |
| Program-of-thoughtModel Name=Gemini2.5-pro2026.02 | 31.8 | 7.4 | 48.9 | 8.8 | 47.8 | 11.7 | 26.2 | 6.9 | 15.6 | 4.5 | |
| Canvas-of-thoughtModel Name=Gemini2.5-pro2026.02 | 31.5 | 7.2 | 49.4 | 8.5 | 48.3 | 9.4 | 26.6 | 6.2 | 18.9 | 5.7 | |
| Tree-of-thoughtModel Name=Gemini2.5-pro2026.02 | 30.4 | 8.2 | 48.3 | 9.1 | 45.9 | 13.7 | 25.6 | 7.3 | 13.5 | 5.2 | |
| Chain-of-thoughtModel Name=Gemini2.5-pro2026.02 | 30 | 1.3 | 47.7 | 1.4 | 46.5 | 1.6 | 25.1 | 0.9 | 12.7 | 1.3 | |
| Iterative ReflectionModel Name=Gemini2.5-pro2026.02 | 25.8 | 7.3 | 42.1 | 8.5 | 40.1 | 11.6 | 20.5 | 6.7 | 10.9 | 4.4 | |
| Chain-of-thoughtModel Name=GPT-5-mini2026.02 | 25.2 | — | 49.4 | — | 19.1 | — | 16.9 | — | 15.3 | — | |
| Program-of-thoughtModel Name=OpenAI GPT-52026.02 | 23.5 | 2.8 | 42.1 | 3.4 | 19.8 | 4.8 | 20.5 | 2.2 | 16 | 1.8 | |
| Tree-of-thoughtModel Name=OpenAI GPT-52026.02 | 20.2 | 6.1 | 34.1 | 6.5 | 17.2 | 8.7 | 16.4 | 5.1 | 15.6 | 5 | |
| Chain-of-thoughtModel Name=OpenAI GPT-52026.02 | 17.4 | 0.7 | 30.7 | 0.8 | 15.3 | 1.5 | 12.3 | 0.5 | 13.8 | 0.3 | |
| Chain-of-thoughtModel Name=Doubao-1.5-vision-pro2026.02 | 15.6 | — | 30.1 | — | 8.9 | — | 12.8 | — | 12 | — | |
| Iterative ReflectionModel Name=OpenAI GPT-52026.02 | 12.7 | 2.6 | 22.7 | 3 | 12.1 | 4.5 | 8.7 | 2.1 | 9.5 | 1.7 | |
| Chain-of-thoughtModel Name=OpenAI GPT-4.52026.02 | 12.6 | — | 18.2 | — | 2.5 | — | 11.8 | — | 15.3 | — |