Tencent Releases Advanced AI Model Hy3 Following Key Hire

Tencent has announced a major update to its Hunyuan AI model, dubbed Hy3, after bringing on former OpenAI scientist Yao Shunyu. The new iteration demonstrates improved performance across various tasks including complex reasoning and coding.

The move signals Tencent’s intensified efforts in the competitive Chinese AI landscape where it faces rivals like ByteDance, Alibaba, and DeepSeek. While still catching up to these competitors, Hy3 represents a significant step forward for Tencent’s AI capabilities.

Key Features of Hy3

• Enhanced performance in complex reasoning tasks
• Improved coding abilities
• Utilizes Yao Shunyu’s expertise from OpenAI

Tencent previously asserted that its Hunyuan LLM was more powerful and intelligent than both ChatGPT and Llama when released in 2023. The company has been actively investing in AI startups like Moonshot AI and StepFun to bolster its cloud computing competitiveness.

Expanding Investments in AI

Tencent plans to double its annual AI investment to over $5 billion (approximately 6.8 trillion won), focusing on improving data quality through organizational restructuring.

This comes as China ramps up open-source AI initiatives to compete with major US tech companies. Meanwhile, DeepSeek recently launched its V4 Flash and V4 Pro LLMs featuring improved reasoning capabilities and a new hybrid attention architecture that allows the models to better retain context in extended conversations.