General Visual Question Answering on RealWorldQA (Accuracy)
71.9AccuracyBARD-VL
Evaluation Results
| Method | Links | |
|---|---|---|
| BARD-VLParameters=4B, B=322026.04 | 71.9 | |
| BARD-VLParameters=8B, B=42026.04 | 70.7 | |
| Qwen3-VLParameters=4B2026.04 | 70.5 | |
| Qwen3-VLParameters=8B2026.04 | 69.5 | |
| GR3D-8BTraining Stage=Stage 12026.05 | 68.75 | |
| Dream-VLParameters=7B2026.04 | 67.7 | |
| Qwen3-VL-8BFine-tuning setting=Base2026.07 | 66.93 | |
| InternVL3.5Parameters=4B2026.04 | 66.7 | |
| SDAR-VLParameters=8B2026.04 | 66.1 | |
| NVILA-Lite-8B2026.05 | 65.35 | |
| GR3D-8BTraining Stage=Stage 22026.05 | 65.23 | |
| Qwen3-VL-8BFine-tuning setting=LoRA2026.07 | 64.84 | |
| BARD-VLParameters=2B, B=322026.04 | 64.6 | |
| LLaDA-VParameters=8B2026.04 | 63.4 | |
| Qwen3-VL-8BFine-tuning setting=CPO2026.07 | 63.27 | |
| InternVL3.5Parameters=8B2026.04 | 63.1 | |
| AVLM-2BModel Scale=2B2026.06 | 62.75 | |
| Qwen3-VL-8BFine-tuning setting=GSPO2026.07 | 59.48 | |
| Qwen2.5-Omni-3BModel Scale=3B2026.06 | 58.69 | |
| Qwen3-VL-8BFine-tuning setting=FFT2026.07 | 55.29 | |
| Dimple-VLParameters=7B2026.04 | 51.9 | |
| Gemma-4-E2B-itModel Scale=4B2026.06 | 43.92 | |
| LaviDaParameters=8B2026.04 | 40.3 | |
| MMaDAParameters=8B2026.04 | 28.2 | |
| Qwen3-VL-8BFine-tuning setting=GRPO2026.07 | 27.45 |