加载中...
加载中...
Hunyuan-T1
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
Hunyuan-T1 currently shows benchmark results led by MMLU Pro (9 / 114, score 87.20), MATH-500 (19 / 43, score 96.20), LiveCodeBench (53 / 106, score 64.90). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.
Hunyuan-T1大模型是腾讯发布的自研大模型,基于其自研的TurboS架构,采用Hybrid-Transformer-Mamba MoE混合设计,结合了多种技术优势:
根据DataLearnerAI收集的大模型评测排行,Hunyuan-T1的推理能力还可以,但是不是业界最强的:

Hunyuan-T1通过混合架构设计与强化学习优化,在推理效率与长文本处理能力上实现了技术突破。其评测表现显示出在复杂任务中的竞争力,尤其在中文场景与特定垂直领域具备应用潜力。然而,模型参数量、多语言扩展性等细节仍有待进一步公开验证。该模型目前通过腾讯“元宝APP”提供服务,暂未公布开源计划。
欢迎关注 DataLearner 官方微信,获得最新 AI 技术推送
