ChainCatcher report: Tencent Hunyuan has officially open-sourced Hy4 Flash, just two days after the model’s release. Three model weight versions are being released at once: Hy4 Flash is the instruction-tuned production model, ready for direct use in dialogue, reasoning, and agent tasks; Hy4 Flash-FP8 is an 8-bit quantized version designed primarily to reduce VRAM usage during deployment; and Hy4 Flash Base is the pre-trained base model without instruction tuning, enabling developers to continue training and customizing it. Hy4 Flash has a total of 224B parameters, with 12B activated per inference. Even after FP8 quantization, deployment requirements remain high. Tencent has made the model available for download on GitHub, Hugging Face, ModelScope, and GitCode; the community has also created a GGUF-quantized version for easier execution across diverse hardware environments. The model is licensed under the Tencent Hy License, permitting developers to use and modify weights under agreed terms, with certain conditions attached; large-scale model API providers that meet specified revenue or user-scale thresholds must obtain separate authorization. As for the flagship 770B-parameter Hy4 model, Tencent states it is still accelerating its open-source release, though no specific timeline has been announced.
Tencent HuanYuan Open-Sources Hy4 Flash with Three Model Versions
ChaincatcherShare
Tencent HuanYuan released and open-sourced the Hy4 Flash model just two days after its launch, offering three versions: Hy4 Flash, Hy4 Flash-FP8, and Hy4 Flash Base. The model has 224B parameters and activates 12B per inference. The FP8 version reduces GPU memory usage but still requires substantial resources. Hy4 Flash is available on GitHub, Hugging Face, and other platforms, with community-provided GGUF quantized versions. The Tencent HuanYuan License permits use and modification, but large API providers must obtain additional authorization if they reach specified revenue or user thresholds. The 770B-parameter Hy4 flagship model is still in the process of being open-sourced, with no timeline provided. This update brings fresh on-chain and crypto news for developers and AI enthusiasts.
Source:Show original
Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information.
Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.