腾讯云 TI-ONE 训练平台

You use Tencent Cloud TI-ONE to run and work with Hunyuan-Large, Tencent's open source MoE large language model, from a cloud training platform.

Tencent Cloud TI-ONE is a training platform for large language models that gives you access to that model for text generation, writing and editing, and code-related work.

Hunyuan-Large is built on a Transformer architecture with a mixture-of-experts design.

In the comparisons Tencent publishes, it is set against DeepSeek-V2, Llama3.1-70B, Llama3.1-405B and Mixtral-8x22B.

The benchmarks named are CMMLU, MMLU and CEval, and results are reported across NLP tasks in nine capability areas.

See vendor benchmarks as a starting point and test the model on your own material before committing.

The listing sits across AI models, open source models and code generation, so it suits developers and teams who want to train or fine-tune a large Chinese-developed model on managed infrastructure rather than set up their own.

Documentation and the model pages are published in Chinese, which matters if you read only English.