A small reasoning vision-language model fine-tuned from Qwen3-VL-2B for physical common sense, such as judging object interactions, spatial relations, and whether a video is physically plausible.
vikhyatk/moondream2
Indexed from Hugging FaceLinked to its home on Hugging Face. FrontierOS indexes it; nothing is hosted or mirrored here.ModelVision-language1.9B paramsEst. Edge GPUEstimated minimum compute for inference: Jetson Orin class, 8–16 GB shared memory. Not a benchmark.Apache 2.0Desk deviceMobile baseSingle arm
About
Compact, frequently updated vision-language model with built-in captioning, visual Q&A, object detection and pointing.
Why it matters for physical AI
Pointing and detection outputs turn a text request into image coordinates a robot can act on.
Summary written by FrontierOS in our own words. The upstream card and license govern use. Metadata checked 27 Sep 2026. Read the full card on Hugging Face
Files, weights and version history live on Hugging Face. FrontierOS indexes this repo; it does not host or mirror it.
hf download vikhyatk/moondream2Uses the Hugging Face CLI (pip install -U huggingface_hub). This downloads the whole repo; add --include "<pattern>" to fetch only the files you need.
More vision-language
Browse allHalf-billion-parameter vision-language model from Hugging Face that answers questions about images and short video clips.
Smallest instruction-tuned model in Alibaba's Qwen3-VL line, covering image and video understanding, OCR and grounded object localization.