# Reasoning under Vision: Understanding Visual-Spatial Cognition in Vision-Language Models for CAPTCHA

arXiv: https://arxiv.org/abs/2510.06067 · HTML: https://mlsanity.com/paper/2510.06067/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Spatial and embodied reasoning, one of 17 directions in LLM reasoning.
- Builds on 8 of its 26 references that are in the map.
- First posted on arXiv 7 Oct 2025.

## Placed in

- LLM reasoning / [Spatial and embodied reasoning](https://mlsanity.com/lr/spatial-and-embodied-reasoning/)
- Multimodal LLMs / vision-language models / [Embodied and 3D vision-language models](https://mlsanity.com/mllm/embodied-and-3d-vision-language-models/)

## Builds on (most cited first, 8 of 8)

- [Do As I Can, Not As I Say: Grounding Language in Robotic Affordances](https://mlsanity.com/paper/2204.01691/) (2022)
- [On the Measure of Intelligence](https://mlsanity.com/paper/1911.01547/) (2019)
- [3DSRBench: A Comprehensive 3D Spatial Reasoning Benchmark](https://mlsanity.com/paper/2412.07825/) (2024)
- [EgoTextVQA: Towards Egocentric Scene-Text Aware Video Question Answering](https://mlsanity.com/paper/2502.07411/) (2025)
- [Omnia de EgoTempo: Benchmarking Temporal Understanding of Multi-Modal LLMs in Egocentric Videos](https://mlsanity.com/paper/2503.13646/) (2025)
- [PTR: A Benchmark for Part-based Conceptual, Relational, and Physical Reasoning](https://mlsanity.com/paper/2112.05136/) (2021)
- [Oedipus: LLM-enchanced Reasoning CAPTCHA Solver](https://mlsanity.com/paper/2405.07496/) (2024)
- [MCA-Bench: A Multimodal Benchmark for Evaluating CAPTCHA Robustness Against VLM-based Attacks](https://mlsanity.com/paper/2506.05982/) (2025)

## Built on by (most cited first, 4 of 4)

- [Prompt-based Adaptation in Large-scale Vision Models: A Survey](https://mlsanity.com/paper/2510.13219/) (2025)
- [COGNITION: From Evaluation to Defense against Multimodal LLM CAPTCHA Solvers](https://mlsanity.com/paper/2512.02318/) (2025)
- [CAPTCHA Solving for Native GUI Agents: Automated Reasoning-Action Data Generation and Self-Corrective Training](https://mlsanity.com/paper/2603.23559/) (2026)
- [CaptchaArena: A Large-Scale, Fine-Grained Dataset for Training Computer-Use Agents on Interactive CAPTCHAs](https://mlsanity.com/paper/2609.31957/) (2026)

## Abstract (arXiv)

CAPTCHA, originally designed to distinguish humans from robots, has evolved into a real-world benchmark for assessing the spatial reasoning capabilities of vision-language models. In this work, we first show that step-by-step reasoning is crucial for vision-language models (VLMs) to solve CAPTCHAs, which represent high-difficulty spatial reasoning tasks, and that current commercial vision-language models still struggle with such reasoning. In particular, we observe that most commercial VLMs (e.g., Gemini, Claude, GPT, etc.) fail to effectively solve CAPTCHAs and thus achieve low accuracy (around 21.9 percent). However, our findings indicate that requiring the model to perform step-by-step reasoning before generating the final coordinates can significantly enhance its solving accuracy, underscoring the severity of the gap. To systematically study this issue, we introduce CAPTCHA-X, the first real-world CAPTCHA benchmark with reasoning, covering seven categories of CAPTCHAs (such as Gobang, hCaptcha, etc.) with step-by-step action solutions and grounding annotations. We further define five reasoning-oriented metrics that enable a comprehensive evaluation of models reasoning capabilities. To validate the effectiveness of reasoning, we also propose a general agentic VLM-based framework that incorporates the models inherent reasoning abilities. Our method achieves state-of-the-art performance across five high-difficulty CAPTCHA types, with an average solving accuracy of 83.9 percent, substantially surpassing existing baselines. These results reveal the limitations of current models and highlight the importance of reasoning in advancing visual-spatial challenges in the future.