Distortion Severity Prediction on PANDABENCH Easy
59AccuracyPANDA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| PANDAModel Type=Open-source2026.04 | 59 | 61 | 58 | 59 | |
| DepictQAModel Type=Open-source, Trained on PANDASET=true2026.04 | 55 | 53 | 45 | 48 | |
| GPT-5 MiniModel Type=Closed-source2026.04 | 36 | 32 | 31 | 29 | |
| GPT-4oModel Type=Closed-source2026.04 | 33 | 34 | 29 | 27 | |
| SeagullModel Type=Open-source2026.04 | 32 | 34 | 26 | 26 | |
| Gemma-3Model Type=Open-source, Parameters=27B2026.04 | 30 | 31 | 30 | 27 | |
| GPT-5 NanoModel Type=Closed-source2026.04 | 29 | 28 | 27 | 21 | |
| Gemini 2.5 ProModel Type=Closed-source2026.04 | 29 | 32 | 25 | 26 | |
| Attentive ProbeModel Type=Baseline2026.04 | 29 | 26 | 27 | 26 | |
| DepictQAModel Type=Open-source2026.04 | 27 | 12 | 20 | 11 | |
| Linear ProbeModel Type=Baseline2026.04 | 27 | 25 | 26 | 15 | |
| RandomModel Type=Baseline2026.04 | 25 | 25 | 25 | 25 | |
| Q-InsightModel Type=Open-source2026.04 | 19 | 19 | 24 | 11 |