> ## Documentation Index > Fetch the complete documentation index at: https://docs.vlm.run/llms.txt > Use this file to discover all available pages before exploring further. # Toolsets > The vision tools Orion runs, grouped by the media you send. Orion runs real vision tools, not a chat model's guess. Pick the toolset by the media you have. Orion-1 calls one tool per model round-trip. Orion-2 imports the same tools in code, for example `vlmrun.image.detect(image, "person")`. See [Code Execution](/agents/code-execution). ## Image Captions and tags for images. Bounding boxes and confidence scores for objects, faces, and people. Pixel-level masks for objects, regions, and features. Key points and structural features, with sub-pixel accuracy. From a text prompt, a sketch, or an existing image. Buttons and interactive components in screenshots. Crop, rotate, upscale, and colorize. ## Documents Layout elements with bounding boxes and reading order. Map text in a document to its location on the page. Correlate pages and documents in one request. ## Video Topic, summary, timestamped chapters, and content tags. Generate a video from a text prompt. Trim, sample, and extract video segments. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.