Tencent releases Hy4 open-weight model with vision under Apache 2.0
Overview
Tencent has released its Hy4 model under the Apache 2.0 license, with official weights, an official FP8 version and a mixed GGUF quantized version.
MJ Kayum says the model has 770B total parameters, 49B active, and a native 1M context window, and that the mixed GGUF quant cuts the footprint from about 1.5TB to about 214GiB. Tencent says practical scores remain close to BF16.
A South China Morning Post report says Hy4 is Tencent's first model with visual understanding, able to process images, analyse charts and perform visual mathematical reasoning. Tencent says it is among the strongest open-weight models available, but acknowledges it trails leading closed-source systems on research-level mathematics, long-horizon coding, model self-improvement and advanced visual question answering. Tencent Hunyuan's own post also calls Hy4 a leading open model with vision.
Written by AI from the articles below · updated Nov 4, 2:57 AM ET
Check the sources:
Developments
2 developments
- Nov 4, 2:01 AM ET · 2 articlesTencent releases Hy4 open-weight model with visual understandingSouth China Morning Post · Tech: Tencent releases Hy4, its first open-weight model with visual understanding
- Oct 22, 9:13 AM ET · 1 articleTencent releases Hy4 preview under Apache 2.0 with official weights, FP8 and GGUF quantsX search: AI availability posts (now available, public beta): Tencent releases Hy4 preview under Apache 2.0 with FP8 and GGUF variants
Article timeline
The articles in this story. Times are ET.
Tencent Hy@TencentHunyuanOfficialTencent Hunyuan introduces Hy4, an open model with visionAITencent Hunyuan introduces Hy4, a new open model with vision capabilities. The company says Hy4 is among the leading open models and is built with cyber and biosecurity in mind. Tencent points to productivity gains in finance, frontend, game development, officework, and research.

- South China Morning Post · TechNewsTencent releases Hy4, its first open-weight model with visual understanding
AITencent has released Hy4, its latest flagship large language model, which adds visual understanding for the first time. According to the company, Hy4 can process images, analyse complex charts and perform visual mathematical reasoning. Tencent says it is one of the strongest open-weight models available today, but acknowledges it still trails leading closed-source systems on research-level mathematics, long-horizon coding, model self-improvement and advanced visual question answering.
MJ Kayum@mjkayumXTencent releases Hy4 preview under Apache 2.0 with FP8 and GGUF variantsAITencent's Hy4 preview is now available under the Apache 2.0 license, with official weights, an official FP8 version, and a mixed GGUF version. The model has 770B total parameters, 49B active, and native 1M context. According to the post, the mixed quant reduces the footprint from about 1.5TB to about 214GiB, and Tencent reports practical scores remain close to BF16.
Heat Trend
Not enough continuous observations to show a trend yet.