Multimodal Search on BCVL
64.1AccuracyGemini-3.1-Pro
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Gemini-3.1-ProEvaluation Mode=Agentic Workflow2026.05 | 64.1 | 2 | |
| HyperEyes-235BTraining Stage=RL, Evaluation Mode=Agentic Workflow2026.05 | 60 | 2.2 | |
| HyperEyes-30BTraining Stage=RL, Evaluation Mode=Agentic Workflow2026.05 | 57.9 | 2.6 | |
| REDSearchEvaluation Mode=Multimodal Deep Search Agents2026.05 | 57.2 | — | |
| HyperEyes-235BTraining Stage=SFT, Evaluation Mode=Agentic Workflow2026.05 | 54.4 | 3.7 | |
| VDREvaluation Mode=Multimodal Deep Search Agents2026.05 | 53.7 | 11.7 | |
| Gemini-3.1-ProEvaluation Mode=Direct Answer2026.05 | 53.1 | — | |
| Kimi-K2.5Evaluation Mode=Agentic Workflow2026.05 | 50.3 | 5.1 | |
| Claude-Opus-4.6Evaluation Mode=Agentic Workflow2026.05 | 48.3 | 2.4 | |
| HyperEyes-30BTraining Stage=SFT, Evaluation Mode=Agentic Workflow2026.05 | 47.6 | 3.9 | |
| Claude-Opus-4.6Evaluation Mode=Direct Answer2026.05 | 43.5 | — | |
| Qwen3-VL-235BEvaluation Mode=Agentic Workflow2026.05 | 37.9 | 2.7 | |
| Qwen3-VL-30BEvaluation Mode=Agentic Workflow2026.05 | 29 | 4.4 | |
| Kimi-K2.5Evaluation Mode=Direct Answer2026.05 | 27.6 | — | |
| WebWatcherEvaluation Mode=Multimodal Deep Search Agents2026.05 | 27 | 4.9 | |
| DeepEyes-V2Evaluation Mode=Multimodal Deep Search Agents2026.05 | 24.8 | 4.3 | |
| Qwen3-VL-235BEvaluation Mode=Direct Answer2026.05 | 21.8 | — | |
| MMSearch-R1Evaluation Mode=Multimodal Deep Search Agents2026.05 | 19.1 | 1.7 | |
| Qwen3-VL-30BEvaluation Mode=Direct Answer2026.05 | 17.2 | — |