Multimodal Search on FVQA
84AccuracyGemini-3.1-Pro
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Gemini-3.1-ProEvaluation Mode=Agentic Workflow2026.05 | 84 | 1.3 | |
| HyperEyes-235BTraining Stage=RL, Evaluation Mode=Agentic Workflow2026.05 | 81.4 | 1.5 | |
| HyperEyes-235BTraining Stage=SFT, Evaluation Mode=Agentic Workflow2026.05 | 80.3 | 1.9 | |
| HyperEyes-30BTraining Stage=RL, Evaluation Mode=Agentic Workflow2026.05 | 79.3 | 1.7 | |
| Kimi-K2.5Evaluation Mode=Agentic Workflow2026.05 | 76.5 | 2.5 | |
| HyperEyes-30BTraining Stage=SFT, Evaluation Mode=Agentic Workflow2026.05 | 76.1 | 2 | |
| Claude-Opus-4.6Evaluation Mode=Agentic Workflow2026.05 | 74.5 | 1.3 | |
| VDREvaluation Mode=Multimodal Deep Search Agents2026.05 | 74.2 | 12.7 | |
| Qwen3-VL-235BEvaluation Mode=Agentic Workflow2026.05 | 70.2 | 1.7 | |
| WebWatcherEvaluation Mode=Multimodal Deep Search Agents2026.05 | 64.3 | 4 | |
| Gemini-3.1-ProEvaluation Mode=Direct Answer2026.05 | 62.7 | — | |
| DeepEyes-V2Evaluation Mode=Multimodal Deep Search Agents2026.05 | 60.6 | 2.8 | |
| Claude-Opus-4.6Evaluation Mode=Direct Answer2026.05 | 60.1 | — | |
| Kimi-K2.5Evaluation Mode=Direct Answer2026.05 | 59.6 | — | |
| MMSearch-R1Evaluation Mode=Multimodal Deep Search Agents2026.05 | 58.4 | 1.3 | |
| Qwen3-VL-30BEvaluation Mode=Agentic Workflow2026.05 | 58 | 2 | |
| Qwen3-VL-235BEvaluation Mode=Direct Answer2026.05 | 44.2 | — | |
| Qwen3-VL-30BEvaluation Mode=Direct Answer2026.05 | 36.7 | — |