Tencent's Hy3 Bets On AI Agents Over Model Size

Tencent launched Hy3, the third generation of its flagship large language model, last week, positioning it as optimized for real-world AI agents rather than benchmark supremacy. The model features a Mixture-of-Experts architecture with 295 billion total parameters, 21 billion activated parameters, and a 256K context window. Tencent said Hy3 is designed for coding assistants and enterprise productivity workflows. Independent evaluations by AI consultancy Flowtivity show Hy3 scoring 84.2 on BrowseComp and 79.1 on the public MCP-Atlas set, competitive with proprietary models like Claude Opus 4.8 and GPT-5.5, with a hallucination rate of 5.4%. On SWE-bench Verified, Hy3 scores 78%, trailing GLM-5.2 at 84.2%. Tencent describes the development philosophy as "Co-Design," where models and AI-native applications evolve together through products including WorkBuddy, Yuanbao, ima, Marvis, and CodeBuddy. The company reports WorkBuddy's internal task success rate increased from 72% to 90% while average execution time fell by 34%. Pricing is set at approximately $0.18 per million input tokens and $0.59 per million output tokens via Tencent Cloud. An FP8-quantized variant fits on a single 8x H200 node. Tencent claims daily token consumption of Hy3 has increased twenty-fold since preview release.