# Know When to Zoom: How Small Is Too Small for Vision-Language Models?

DevFeed: [Know When to Zoom: How Small Is Too Small for Vision-Language Models?](<https://devfeed.tech/articles/know-when-to-zoom-how-small-is-too-small-for-vision-language-models-140070.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.22709>)

Author: Siddharth Vohra, Min Xu

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [vlm](<https://devfeed.tech/topics/vlm.md>), [image recognition](<https://devfeed.tech/topics/image-recognition.md>)

Tags: [computer-vision](<https://devfeed.tech/tags/computer-vision.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [vision](<https://devfeed.tech/tags/vision.md>), [vision-language-models](<https://devfeed.tech/tags/vision-language-models.md>), [zoom](<https://devfeed.tech/tags/zoom.md>)

## AI overview

Online vision-language services can shrink or reject large images before the model sees them, making small details unreadable. Across seven models from four vendors, the study finds that the smallest text read half the time rises from 6 pixels or less on small images to a predicted 10-13 pixels on large images at default settings. Tiling images into up to nine crops improves Claude and Gemini accuracy by 11.1-14.5 percentage points and raises GPT-6.1 Sol from 50.7% to 83.7% by avoiding rejections. A two-step zoom, where the model identifies a region before inspecting it, matches or beats tiling on Gemini and comes within 2.5 points on Claude using about 40% of the image tokens. Claude and Gemini both often admit uncertainty when detail is lost, but Gemini is more likely to answer confidently and incorrectly. On Claude, tiling only after an unsure initial reply nearly matches always tiling while using 62-70% of its image tokens.

## Source excerpt

arXiv:2611.22709v1 Announce Type: cross Abstract: Online vision-language models often miss small details in large images, because the service shrinks the image, or rejects it, before the model ever sees it. A car 20 pixels wide in an 8000 by 6000 drone photo reaches Claude about 6 pixels wide, and OpenAI's default rejects the photo outright. We measure how much detail survives for seven models from four vendors, using an eye exam of printed codes and 601 questions about small objects in real photos. The smallest text a model reads half the time grows from 6 pixels or less on a small image to a predicted 10 to 13 pixels on a 7680 by 4320 photo at default settings. Zooming wins much of the detail back. Splitting each photo into up to nine tiles raises accuracy by 11.1 to 14.5 percentage points for Claude and Gemini, and by avoiding rejections it lifts GPT-6.1 Sol from 50.7% to 83.7%. A two-step zoom, in which the model first names the region it needs, matches or beats tiles on Gemini and comes within 2.5 points of them on Claude, with about 40% of their image tokens. When a detail is lost, Claude and Gemini both say they cannot tell about half the time, but they fail differently in the rest: Gemini gives a confident wrong answer in 43% to 48% of these cases, Claude in only 5% to 9%. A check based on detail size, run before the call, misses most of the gain from tiles on Claude and Gemini, but Claude's own reply is a reliable trigger: tiling only after an unsure reply comes within 1.4 points of always tiling with 62% to 70% of the image tokens. We turn these findings into five rules for when an agent should zoom, and release the eye exam, every model answer and code at https://siddvoh.com/know-when-to-zoom.