Tencent releases WeVisOne-VL, Apache 2.0 vision language model in 2B and 4B sizes
Tencent's WeChat Vision team released WeVisOne-VL, an Apache 2.0 vision language model in 2B and 4B sizes. One model handles detection, OCR, counting, segmentation and alpha matting, using a 9.6M-parameter pixel decoder.
Tencent's WeChat Vision team released WeVisOne-VL, a vision language model licensed under Apache 2.0 and offered in 2B and 4B sizes, according to a MarkTechPost report.
What the model does
WeVisOne-VL handles several vision tasks in a single model:
- Detection
- OCR
- Counting
- Segmentation
- Alpha matting
The model includes a 9.6M-parameter pixel decoder.
Key specifications
| Item | Detail |
|---|---|
| Developer | Tencent (WeChat Vision team) |
| Model sizes | 2B and 4B |
| License | Apache 2.0 |
| Pixel decoder | 9.6M parameters |
| Tasks | Detection, OCR, counting, segmentation, alpha matting |
What this means
The Apache 2.0 license places no usage restrictions beyond its standard terms, which matters for teams that want to deploy or modify the weights. The 2B and 4B sizes are small, and covering detection, OCR, counting, segmentation and alpha matting in one model means these tasks can be served by a single checkpoint instead of separate task-specific models.
Related Articles
Tencent Hunyuan releases Hyra 2.0, a research agent that builds reusable tools, under Apache 2.0
Tencent Hunyuan released Hyra 2.0 on November 2, a multi-agent research system designed to build reusable research tools rather than solve single problems. Results and code are published in an Apache 2.0 GitHub repository, and all figures are Tencent's own.
Xiaomi open-sources Xiaomi-OCR-0, a 0.8B document model scoring 96.83 on OmniDocBench v1.6
Xiaomi has released the weights and inference code for Xiaomi-OCR-0, a 0.8B-parameter model that handles document parsing, key information extraction and OCR visual question answering. Xiaomi reports 96.83 on OmniDocBench v1.6, first among models under 1B parameters and second overall.
Sarvam AI releases Prism, a 105B-based decision model that claims 44 ms responses and a 71.8 average score
Sarvam AI has launched Prism, a System 1 decision model built as a LoRA fine-tune of Sarvam 105B. The company claims 44 ms response times and a 71.8 average score across three benchmarks. Weights are on Hugging Face under Apache 2.0.
Tencent releases Hy4 with 1.0M context and first visual understanding, priced at $0.417 / $1.251 per 1M tokens
Tencent has released the official version of Hy4, with a 1.0M-token context window and visual understanding for the first time. It is listed on OpenRouter at $0.417 input / $1.251 output per 1M tokens, shown as 50% off.
Comments
Loading...