Tencent Launches Hunyuan Hy3, Boosting Post-Training Data and RL Compute
MT Newswires Live
Jul 06
Tencent released Hunyuan Hy3, an upgraded preview-to-production model that improves post-training data quality and diversity and expands reinforcement-learning compute. Tencent says Hy3 delivers marked gains on inference, agent and long-context tasks, producing results comparable to larger domestic and international flagship models that are typically 2–5x its parameter size. Hy3 is open-sourced under Apache 2.0 on GitHub, HuggingFace, ModelScope and GitCode.
Tencent also cut Hy3 API and product usage costs on Tencent Cloud and in products such as WorkBuddy. Hy3 API pricing: 1 yuan per mln input tokens, 4 yuan per mln output tokens; cached input tokens 0.25 yuan per mln.
Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.