x.comthreadTrunk Monkey (@C4ndide) on X
NVIDIA released a 3-billion-parameter vision-language model specialized for fast, accurate visual grounding, using parallel box decoding to run roughly 10x faster than Qwen3-VL. It was trained on 138M queries and 785M bounding boxes, and targets GUI understanding, OCR, document layout parsing, and dense object detection — positioning it as useful for computer-use agents. The model is open source.
TL;DR NVIDIA open-sources a 3B VLM for real-time visual grounding, 10x faster than Qwen3-VL, aimed at GUI/OCR/computer-use tasks.
Organisations named here
Technologies named here
Techniques named here
Related in multimodal
latent.spacearticle[AINews] Pi 1.0, Pi Durable, and AIE NYC
github.comrepoShow HN: Made an open-source Lego AI generator
simonwillison.nettool