# Efficient Test-Time Scaling for Small Vision-Language Models

arXiv: https://arxiv.org/abs/2510.03574 · HTML: https://mlsanity.com/paper/2510.03574/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 15 of its 87 references that are in the map.
- First posted on arXiv 3 Oct 2025.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)
- Multimodal LLMs / vision-language models / [Prompt learning and adaptation of CLIP](https://mlsanity.com/mllm/prompt-learning-and-adaptation-of-clip/)

## Builds on (most cited first, 15 of 15)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Qwen Technical Report](https://mlsanity.com/paper/2309.16609/) (2023)
- [Gemma: Open Models Based on Gemini Research and Technology](https://mlsanity.com/paper/2403.08295/) (2024)
- [Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scaling](https://mlsanity.com/paper/2501.17811/) (2025)
- [Towards a Human-like Open-Domain Chatbot](https://mlsanity.com/paper/2001.09977/) (2020)
- [AMBER: An LLM-free Multi-dimensional Benchmark for MLLMs Hallucination Evaluation](https://mlsanity.com/paper/2311.07397/) (2023)
- [A Survey on Test-Time Scaling in Large Language Models: What, How, Where, and How Well?](https://mlsanity.com/paper/2503.24235/) (2025)
- [Efficient Guided Generation for Large Language Models](https://mlsanity.com/paper/2307.09702/) (2023)
- [Ovis2.5 Technical Report](https://mlsanity.com/paper/2508.11737/) (2025)
- [Efficient Test-Time Scaling via Self-Calibration](https://mlsanity.com/paper/2503.00031/) (2025)
- [Beyond Model Adaptation at Test Time: A Survey](https://mlsanity.com/paper/2411.03687/) (2024)
- [LLMs Can Generate a Better Answer by Aggregating Their Own Responses](https://mlsanity.com/paper/2503.04104/) (2025)
- [PlanGEN: A Multi-Agent Framework for Generating Planning and Reasoning Trajectories for Complex Problem Solving](https://mlsanity.com/paper/2502.16111/) (2025)
- [Adapting Vision-Language Models Without Labels: A Comprehensive Survey](https://mlsanity.com/paper/2508.05547/) (2025)
- [Self-Evolved Preference Optimization for Enhancing Mathematical Reasoning in Small Language Models](https://mlsanity.com/paper/2503.04813/) (2025)

## Built on by (most cited first, 12 of 12)

- [Zoom in, Click out: Unlocking and Evaluating the Potential of Zooming for GUI Grounding](https://mlsanity.com/paper/2512.05941/) (2025)
- [AVIS: Adaptive Test-Time Scaling for Vision-Language Models](https://mlsanity.com/paper/2606.11576/) (2026)
- [On Test-Time Scaling for Vision-Language Models](https://mlsanity.com/paper/2606.28864/) (2026)
- [Rethinking Test Time Scaling for Flow-Matching Generative Models](https://mlsanity.com/paper/2511.22242/) (2025)
- [Geometric Risk Control for Vision-Language Model OCR](https://mlsanity.com/paper/2603.19790/) (2026)
- [Architectural Sampling: Test-Time Scaling via Computational Diversity in Frozen Vision-Language Models](https://mlsanity.com/paper/2610.01687/) (2026)
- [It's the Decoding Format, Not the Perturbation: Auditing Consistency-Based Selection for Vision-Language Test-Time Scaling](https://mlsanity.com/paper/2608.01207/) (2026)
- [When Does Consensus Mean Correctness? Measuring the Agreement-Accuracy Coupling with Semantics-Preserving Re-Rendering](https://mlsanity.com/paper/2608.05670/) (2026)
- [ReSight-SMC: Two-Stage Power Sampling via Island SMC with Visual Scouts](https://mlsanity.com/paper/2609.34905/) (2026)
- [Token-Disentangled Latent Test-Time Scaling for Vision-Language Reasoning](https://mlsanity.com/paper/2609.35228/) (2026)
- [Harness Compilation: Which Decisions Should a Small Vision-Language Model Keep?](https://mlsanity.com/paper/2610.11231/) (2026)
- [On the Limits of Representation-Based Test-Time Adaptation for Multimodal LLMs](https://mlsanity.com/paper/2610.19892/) (2026)

## Abstract (arXiv)

Small Vision-Language Models (VLMs) provide a computationally efficient alternative to larger models, at the cost of weaker generalization abilities and downstream task performance. These shortcomings could be addressed by test-time scaling techniques, but existing methods are typically computationally demanding, contradicting the resource-efficient design goals of small models. To address these limitations, we propose two novel and efficient test-time scaling strategies that leverage the model-internal features rather than external supervision: (i) Test-Time Augmentation (TTAug), which generates multiple augmented inputs and aggregates outputs at the token level without parameter updates, and (ii) Test-Time Adaptation (TTAdapt), which adapts model parameters during inference using consensus-based pseudolabels from TTAug. Through extensive experiments across nine benchmarks, we demonstrate consistent performance improvements while maintaining computational efficiency suitable for resource-constrained environments. The generality of our approach is demonstrated both within models at different scales and across different VLMs without additional tuning.