JigShape: Evaluating Visual-Geometric Reasoning in VLMs through Jigsaw Puzzles

arXiv:2607.27670v1 Announce Type: new Abstract: Jigsaw puzzle solving requires jointly reasoning about visual content and geometric constraints, yet existing benchmarks use rectangular cuts that create ambiguous ground truth in texture-repeated regions. We introduce extit{ours{}}, a benchmark with tab-and-blank interlocking pieces where geometric constraints provide strong local compatibility requirements that, combined with visual content, yield unambiguous ground truth. Across 95K instances at four grid densities (4imes4 to 16imes16), we find that extbf{zero-shot VLMs largely lack geometric reasoning}: only one of five frontier models (GPT-5.5) exceeds random baseline on 4imes4 puzzles, while all others perform at chance level. While supervised fine-tuning achieves >97% on 4imes4, extbf{all models collapse on larger grids}: GPT-5.5 drops from 70% to near-random on 8imes8, and even fine-tuned models fall below 5% on 12imes12. This ``scaling cliff'' suggests current architectures cannot maintain consistent constraint satisfaction as the number of pieces increases. ours{} establishes scalable geometric reasoning as an open challenge for vision-language models.

Related

• Eliciting Complex Spatial Reasoning in MLLMs through Wide-Baseline Matching
• GeoWeaver: Grounding Visual Tokens with Geometric Evidence before Scene Reasoning
• GeoArena: Evaluating Open-World Geographic Reasoning in Large Vision-Language Models

Source: arXiv cs.CV | 2026-07-31