Infographic Visual Question Answering on InfoVQA (test)
83.1AccuracyQwen3-VL-8B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3-VL-8BParameter Scale=8B2026.05 | 83.1 | |
| Gemini 1.5 Pro2024.12 | 81 | |
| PerceptionLM-8BParameter Scale=8B2026.05 | 80.9 | |
| Qwen3-VL-4BParameter Scale=4B2026.05 | 80.3 | |
| Molmo2-8BParameter Scale=8B2026.05 | 80.1 | |
| GPT-4o2024.12 | 79.2 | |
| InternVL3.5-8BParameter Scale=8B2026.05 | 79.1 | |
| InternVL2# Parameters=40B2024.12 | 78.7 | |
| Molmo2-4BParameter Scale=4B2026.05 | 78.6 | |
| InternVL3.5-4BParameter Scale=4B2026.05 | 78 | |
| Qwen2-VL# Parameters=8B2024.12 | 76.5 | |
| LLaVA-OneVision# Parameters=72B2024.12 | 74.9 | |
| InternVL2# Parameters=8B2024.12 | 74.8 | |
| Zamba2-VL-7BParameter Scale=7B2026.05 | 74.8 | |
| PerceptionLM-3BParameter Scale=3B2026.05 | 74.6 | |
| Claude 3.5 Sonnet2024.12 | 74.3 | |
| NVILA# Parameters=15B2024.12 | 73.5 | |
| Qwen3-VL-2BParameter Scale=2B2026.05 | 72.4 | |
| InternVL3.5-2BParameter Scale=2B2026.05 | 70.8 | |
| NVILA# Parameters=8B2024.12 | 70.7 | |
| NVILA-Lite# Parameters=15B2024.12 | 69.3 | |
| LLaVA-OneVision# Parameters=8B2024.12 | 68.8 | |
| NVILA-Lite# Parameters=8B2024.12 | 67.9 | |
| Zamba2-VL-2.7BParameter Scale=2.7B2026.05 | 66.5 | |
| PerceptionLM-1BParameter Scale=1B2026.05 | 63 | |
| Zamba2-VL-1.2BParameter Scale=1.2B2026.05 | 60.7 | |
| InternVL3.5-1BParameter Scale=1B2026.05 | 60.5 | |
| Florence-VL# Parameters=8B2024.12 | 51.7 | |
| Pixtral# Parameters=12B2024.12 | 50.8 | |
| Cambrian-1# Parameters=34B2024.12 | 46 | |
| Cambrian-1# Parameters=8B2024.12 | 41.6 | |
| VILA-1.5# Parameters=40B2024.12 | 38.4 | |
| VILA-1.5# Parameters=13B2024.12 | 30.4 | |
| LLaVA-1.5# Parameters=13B2024.12 | 29.4 | |
| VILA-1.5# Parameters=8B2024.12 | 25.9 | |
| LLaVA-1.5# Parameters=7B2024.12 | 25.8 | |
| Gemini-1.5-Pro2024.12 | 0.81 | |
| Qwen2-VLSize=7B, #token /image tile=Native resolution (unfixed)2024.12 | 0.765 | |
| PVC InternVL2 (Ours)Size=8B, #token /image tile=642024.12 | 0.75 | |
| InternVL2Size=8B, #token /image tile=2562024.12 | 0.748 | |
| IXC-2.5Size=7B, #token /image tile=4002024.12 | 0.699 | |
| LLaVA-OVSize=7B, #token /image tile=7292024.12 | 0.688 | |
| Qwen2-VLSize=2B, #token /image tile=Native resolution (unfixed)2024.12 | 0.655 | |
| PVC InternVL2 (Ours)Size=2B, #token /image tile=642024.12 | 0.599 | |
| InternVL2Size=2B, #token /image tile=2562024.12 | 0.589 | |
| LongVASize=7B, #token /image tile=1442024.12 | 0.494 |