Mirage is a vision model by Cactus Compute for image understanding and structured extraction, under the Apache License 2.0. Its model card lists scene descriptions, object detection, OCR, segmentation and spatial relations. It runs in the cactus engine; in Python it is one call, cactus.Mirage().describe("photo.png").1
What it returns
The answer is JSON. Each object has a name, a confidence score, a bounding box and an outline, both in coordinates from 0 to 1000. It also gets a position such as "center" and a short description with colour, shape, state and material. Text in the image comes as strings with their boxes, and the scene as a few fields such as setting, lighting or style. The answers are in English. The file mirage.cact is about 21.6 MB, and in my browser test a photo took under a second.2
-
Project notes, 18 November 2026. ↩
- Used in
-
- What the Machine Sees A small app from my old website: choose a photo, and the vision model Mirage found objects, text and the scene in it, entirely in the browser.
- By
-
- Cactus Compute Company behind cactus, an engine for running small AI models on consumer devices, and models such as the vision model Mirage.
- See also
-
- cactus Open-source engine by Cactus Compute that runs small AI models for text, speech and images locally on consumer devices.
- Tags