Tencent Unveils New AI Model Hunyuan Hy3 with Advanced Reasoning Capabilities

Photo: TechNode
Quick answer
Tencent launched the Hunyuan Hy3 AI model featuring a Mixture of Experts architecture, 295B parameters, and 256K token context support.
Chinese tech giant Tencent has unveiled its new Hunyuan Hy3 AI model, which supports advanced reasoning mechanisms. Built on a Mixture of Experts (MoE) architecture, the model features 295 billion parameters, with 21 billion activated during operation. This enables Hy3 to efficiently process large data volumes with a context window of up to 256,000 tokens.
The model is already integrated into several key Tencent products, including automation tools WorkBuddy and CodeBuddy, as well as platforms Yuanbao, Marvis, and ima. API access to Hunyuan Hy3 is available via the Tencent Cloud TokenHub platform, allowing developers to leverage its capabilities in their own solutions.
The API usage cost is set at 1 yuan ($0.15) per million input tokens and 4 yuan ($0.59) per million output tokens. For cached input data, the price is reduced to 0.25 yuan ($0.037) per million tokens, making Hy3 accessible to both corporate and individual users.
Common questions
- What are the key features of the Hunyuan Hy3 model?
- Hunyuan Hy3 is built on a Mixture of Experts (MoE) architecture with 295 billion parameters, of which 21 billion are activated during operation. It supports a context window of up to 256,000 tokens and combines fast and slow reasoning mechanisms.
- Which Tencent products integrate the Hy3 model?
- The Hunyuan Hy3 model is already integrated into Tencent products such as WorkBuddy, CodeBuddy, Yuanbao, Marvis, and ima, enhancing their AI capabilities.
- What is the cost of using the Hunyuan Hy3 API?
- The Hunyuan Hy3 API costs 1 yuan ($0.15) per million input tokens and 4 yuan ($0.59) per million output tokens. Cached input tokens are priced at 0.25 yuan ($0.037) per million.
Dzen feed: /feed/dzen.xml · RSS: /feed.xml