High-Resolution Multimodal Reasoning on HR-Bench 4K FCP
78.3ACCRTWI
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RTWIModel Variant=Qwen3-VL Instruct, Online setting=true2026.02 | 78.3 | 46.4 | |
| SCModel Variant=Qwen3-VL Instruct, Online setting=true2026.02 | 77 | — | |
| ESCModel Variant=Qwen3-VL Instruct, Online setting=true2026.02 | 77 | 23.1 | |
| ASCModel Variant=Qwen3-VL Instruct, Online setting=true2026.02 | 76.8 | 39.9 | |
| DeepconfModel Variant=Qwen3-VL Instruct, Online setting=true2026.02 | 76.8 | 38.3 | |
| CISCModel Variant=Qwen3-VL Instruct, Online setting=true2026.02 | 75.8 | 41.2 | |
| Self-Cer.Model Variant=Qwen3-VL Instruct, Online setting=true2026.02 | 75.8 | 43.2 | |
| BaseModel Variant=Qwen3-VL Instruct, Online setting=true2026.02 | 72 | — | |
| RTWIModel Variant=Qwen3-VL Thinking, Online setting=true2026.02 | 65.8 | 50.5 | |
| GPT-4oOnline setting=true2026.02 | 63.3 | — | |
| ThymeOnline setting=true2026.02 | 63 | — | |
| DeepconfModel Variant=Qwen3-VL Thinking, Online setting=true2026.02 | 62.5 | 48.9 | |
| ASCModel Variant=Qwen3-VL Thinking, Online setting=true2026.02 | 61.8 | 29.3 | |
| SCModel Variant=Qwen3-VL Thinking, Online setting=true2026.02 | 61 | — | |
| ESCModel Variant=Qwen3-VL Thinking, Online setting=true2026.02 | 61 | 18.9 | |
| DeepEyesOnline setting=true2026.02 | 59 | — | |
| CISCModel Variant=Qwen3-VL Thinking, Online setting=true2026.02 | 59 | 42.9 | |
| Self-Cer.Model Variant=Qwen3-VL Thinking, Online setting=true2026.02 | 59 | 45.4 | |
| BaseModel Variant=Qwen3-VL Thinking, Online setting=true2026.02 | 57.3 | — |