Moondream
Open vision models made for production environments, with built-in grounding skills.
Moondream is a family of open vision-language models built for production computer-vision workloads, with variants including Moondream 3/3.1, Moondream 2, and Moondream 2 0.5B. The family supports grounded visual skills such as object detection, pointing, captioning, visual question answering, segmentation, OCR, structured output, and reasoning, and can be deployed through Moondream Cloud or locally with Photon.
Model Intelligence
Context window
32,000 tokens
Benchmarkable
No
Model level
family
Recent stories
0 linked stories
No linked stories yet.