--- title: README emoji: 👁️ colorFrom: blue colorTo: indigo sdk: static pinned: false ---

Demo: 256 images, one request, a scan picks the right image, the view zooms in, and new requests find each target in it

Vision-Decision builds Glance, which answers many questions about one image in one forward pass, returning typed decisions: a box, yes or no, a choice, or not there.

视觉决策模型:一次前向,回答一张图上的许多问题。

**Why.** Agents mostly need decisions, not prose: *Where is the button? Is it there at all? Which clip shows it?* Generative VLMs write every answer token by token, which is slow and turns "not there" into a guess. Glance reads them all from one pass. ## Models | model | what it is | |---|---| | [Glance-4B-preview](https://huggingface.co/Vision-Decision/Glance-4B-preview) | a 4B decision VLM on Qwen3.5-4B: boxes, yes/no, choices and scores, with calibrated "not there", from one prefill | | [Glance-X-4B](https://huggingface.co/Vision-Decision/Glance-X-4B) | bidirectional grounding: reuse the image across requests or the request across images, with multi-box output | | [Glance-X-Lite](https://huggingface.co/Vision-Decision/Glance-X-Lite) | a 0.8B Glance-X for Apple silicon and other small devices | ## Data | dataset | what it is | |---|---| | [Glance-COCO](https://huggingface.co/datasets/Vision-Decision/Glance-COCO) | grounding on COCO images with exact boxes, many phrasings, Chinese requests, hard negatives and multi-target queries; plus training supplements, trainer-format records and the `robust` evaluation set (formerly Glance-Data) | | [Glance-GUI](https://huggingface.co/datasets/Vision-Decision/Glance-GUI) | synthetic GUI grounding: 18.5k screenshots, 1.08M requests with verified hard negatives | | [Glance-Video](https://huggingface.co/datasets/Vision-Decision/Glance-Video) | typed-question records for short video clips (annotations only) | | [Glance-Experiments](https://huggingface.co/datasets/Vision-Decision/Glance-Experiments) | results and model outputs: the Decision Index run, the referring-negative audit, scan-engine measurements | Browse by collection: [Glance models](https://huggingface.co/collections/Vision-Decision/glance-models-6affec89a20ff6e9d31de39f) · [Glance training data](https://huggingface.co/collections/Vision-Decision/glance-training-data-6affec8bf34e26467ed86667) · [Glance evaluation](https://huggingface.co/collections/Vision-Decision/glance-evaluation-6affec8fdeb34f65d8f366c9) · [Collaborator projects](https://huggingface.co/collections/Vision-Decision/collaborator-projects-6affec914cc1ae6192af1651) ## Code [Glance](https://github.com/Vision-Decision/Glance) (inference package) · [Glance-Core](https://github.com/Vision-Decision/Glance-Core) (training, data builders, evaluation) · [Glance-Engine](https://github.com/Vision-Decision/Glance-Engine) (serving engine) · [Glance-Video](https://github.com/Vision-Decision/Glance-Video) (video and clip scanning) · [Glance-X-MLX](https://github.com/Vision-Decision/Glance-X-MLX) · [Glance-COCO](https://github.com/Vision-Decision/Glance-COCO) (data pipeline)