Image Question Answering on Molmo QA Benchmarks Image 19
86.2Image Average AccuracyGemini 3 Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini 3 ProAccess=API call only2026.03 | 86.2 | |
| GPT-5Access=API call only2026.03 | 83.7 | |
| MolmoPoint-8BAccess=MolmoPoint2026.03 | 82.2 | |
| GPT-5 miniAccess=API call only2026.03 | 81.9 | |
| Molmo2-8BAccess=Fully open2026.03 | 81.7 | |
| Gemini 2.5 ProAccess=API call only2026.03 | 81.3 | |
| Qwen3-VL-8BAccess=Open weights2026.03 | 81.2 | |
| Eagle2.5-8BAccess=Open weights2026.03 | 81.2 | |
| Molmo2-4BAccess=Fully open2026.03 | 80.4 | |
| Keye-VL-1.5-8BAccess=Open weights2026.03 | 79.8 | |
| Molmo2-O-7BAccess=Fully open2026.03 | 79.7 | |
| Gemini 2.5 FlashAccess=API call only2026.03 | 79.3 | |
| PLM-8BAccess=Fully open2026.03 | 78.7 | |
| Qwen3-VL-4BAccess=Open weights2026.03 | 78.4 | |
| InternVL3.5-8BAccess=Open weights2026.03 | 78.2 | |
| MiniCPM-V-4.5-8BAccess=Open weights2026.03 | 77.7 | |
| InternVL3.5-4BAccess=Open weights2026.03 | 77.2 | |
| GLM-4.1V-9BAccess=Open weights2026.03 | 77 | |
| Claude Sonnet 4.5Access=API call only2026.03 | 76.3 | |
| PLM-3BAccess=Fully open2026.03 | 75.9 |