List your company

Sliding Puzzle

56.3Solve RateAdaptive (Qwen2.5-VL-7B)
-2.25212.94928.1543.351May 11, 2026

Evaluation results

MethodLinks
Adaptive (Qwen2.5-VL-7B)Group=Ours2026.0556.337
GPT-5.5Group=Closed (zero-shot)2026.0522.235.1
Gemini 3.1 ProGroup=Closed (zero-shot)2026.0511.136.1
Claude SonnetGroup=Closed (zero-shot)2026.050—
InternVL3-8B/14B/78BGroup=Open (zero-shot)2026.050—
Qwen2.5-VL-7B/72BGroup=Open (zero-shot)2026.050—
Qwen3-VL-8B/32BGroup=Open (zero-shot)2026.050—