WeVisOne-VL
by Tencent · Model · Image
unproven
Open vision model for detection, OCR and segmentation; only Tencent's benchmarks so far
Vision-language model for detection, counting, pointing, OCR, segmentation and alpha matting; code and weights released
AccessOpen weightsNeeds~19 GBDownloadabout 19 GB for the 4B model; a 2B variant is about half thatSize4B parametersLicenseApache 2.0
Why we say so
Tencent says it beats competitors across detection, counting, OCR, segmentation and matting, and nobody independent has tested that. AI Search described the 4B model's size and its Qwen3 VL base but did not report results.
Evidence
- Oct 25Shown outperforming competitor models across detection, counting, OCR, segmentation and matting benchmarks
Cost to run
Alternatives
Made from