List your company

Overall Driving Performance on GEODRIVE-BENCH (cross-country standard deviation)

3.47Std Dev of Overall AccuracyQwen2.5-VL-7B
3.12765.43887.7510.0612Jun 1, 2026

Evaluation results

MethodLinks
Qwen2.5-VL-7BSetting=Rule-Given2026.063.47
DriveOPD‡Setting=Reasoning, backbone=InternVL3-8B2026.063.76
InternVL3-8BSetting=Rule-Given2026.064.04
DriveOPD†Setting=Direct, backbone=Qwen2.5-VL-7B2026.064.48
DriveOPD‡Setting=Rule-Given, backbone=InternVL3-8B2026.064.58
DriveOPD‡Setting=Direct, backbone=InternVL3-8B2026.064.89
Qwen3-VL-8BSetting=Rule-Given2026.064.98
DriveOPD†Setting=Reasoning, backbone=Qwen2.5-VL-7B2026.065.14
DriveOPD†Setting=Rule-Given, backbone=Qwen2.5-VL-7B2026.065.56
Llama-3.2-11B-VSetting=Rule-Given2026.065.73
InternVL3.5-8BSetting=Rule-Given2026.065.82
LLaVA-1.6-7BSetting=Rule-Given2026.066.13
Gemma3-12BSetting=Rule-Given2026.066.9
Gemma3-12BSetting=Direct2026.067.79
InternVL3.5-8BSetting=Direct2026.068.68
Qwen3-VL-8BSetting=Reasoning2026.069.87
Gemma3-12BSetting=Reasoning2026.0610.15
LLaVA-1.6-7BSetting=Direct2026.0610.41
LLaVA-1.6-7BSetting=Reasoning2026.0610.48
InternVL3-8BSetting=Direct2026.0610.85
InternVL3.5-8BSetting=Reasoning2026.0610.87
Llama-3.2-11B-VSetting=Direct2026.0610.91
InternVL3-8BSetting=Reasoning2026.0611.4
Qwen3-VL-8BSetting=Direct2026.0611.45
Qwen2.5-VL-7BSetting=Reasoning2026.0611.48
Llama-3.2-11B-VSetting=Reasoning2026.0611.5
Qwen2.5-VL-7BSetting=Direct2026.0612.03