End-to-End Document Question Answering on InfoVQA (test)
86.9Overall ScoreGemini-1.5-Pro
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Gemini-1.5-ProArchitecture=Multimodal LLMs2025.10 | 86.9 | 81.1 | 95.4 | |
| SlideAgentArchitecture=Multimodal RAG and Agentic Methods2025.10 | 79.6 | 69.9 | 94.1 | |
| Gemini-1.5-FlashArchitecture=Multimodal LLMs2025.10 | 72.3 | 66.1 | 81.3 | |
| Gemini-1.5-Flash-LiteArchitecture=Multimodal LLMs2025.10 | 71.7 | 62.6 | 87 | |
| ViDoRAGArchitecture=Multimodal RAG and Agentic Methods2025.10 | 71.2 | 60.5 | 90.7 | |
| GPT-4oArchitecture=Multimodal LLMs2025.10 | 69 | 59.3 | 90.5 | |
| Claude-3-OpusArchitecture=Multimodal LLMs2025.10 | 36.3 | 35.4 | 38.9 | |
| Claude-3.5-HaikuArchitecture=Multimodal LLMs2025.10 | 31.5 | 30.8 | 37 |