正文 · AI 翻译
On November 5, Tencent announced that it had expanded WorkBuddy's capacity tenfold due to a surge in Hy4 calls.
On November 5, Tencent announced that it had expanded WorkBuddy's capacity tenfold due to a surge in Hy4 calls, to meet more users' needs for access and use. Meanwhile, as its model capabilities continue to improve, Hy4 has received strong recognition from authoritative international model rankings, placing second among Chinese models on the Artificial Analysis Intelligence Index and first among open-source models on the Agent Arena leaderboard.
Since Hunyuan Hy4 first became available in WorkBuddy, its significantly improved agent capabilities have prompted many users and developers to try it and offer enthusiastic feedback. On its first day, many users reached their usage limits. Tencent has now expanded the Hy4 inference cluster tenfold, increased users' free usage allowance, and continues to allocate resources dynamically.
The enthusiastic response from users has been matched by strong results for Hy4 in several authoritative overseas evaluations. On November 5, Artificial Analysis, a leading global platform for evaluating large language models, released its latest Intelligence Index. Tencent's Hy4 ranked among the leaders out of 724 models and second among Chinese models, behind only Kimi K3.1 Pro and ahead of models including Qwen 3.8 Max and DeepSeek V4.1 Flash.
Artificial Analysis is regarded within the industry as an important reference for assessing the overall intelligence of large language models. Its index is not a single benchmark score but a demanding evaluation combining core capabilities including complex reasoning, code generation and debugging, tool use, multimodal understanding, long-range planning, and scientific problem-solving, with high requirements for a model's ability to generalize and its practical utility. Hy4 tied with Grok 4.7 (xhigh) in this comprehensive evaluation, reflecting Tencent's solid foundation in base-model capabilities.
Artificial Analysis also noted that, in terms of cost efficiency, Hy4 costs about $0.78 per task, significantly less than Grok 4.7 (xhigh) and Qwen3.8 Max in the same score range. It is 10 times and 13 times cheaper than those two models, respectively, demonstrating a strong balance between capability and cost.
In addition, on the latest Agent Arena leaderboard, which focuses more on performance in real-world agent tasks, Tencent's Hy4 ranked 13th overall and first among open-source models worldwide. Thanks to its relatively strong value for money, Hy4 also sits on the leaderboard's Pareto Frontier, maintaining a positive score while costing significantly less than top models such as Claude Opus 5.5.
On November 4, Tencent released Hy4, the latest generation of its Hunyuan large language model, with another broad improvement in capabilities. Compared with Hy4 preview, Hy4 significantly expanded the data and scale of both pre-training and post-training, introduced visual understanding capabilities for the first time, and substantially improved its overall performance in instruction following, long-text comprehension, and long-range agent tasks.
This article was provided by Tencent and republished by QbitAI with permission. The views expressed are those of the original author.