Skip to content

Tencent-Hunyuan-Large

An open MoE model with 389 billion total parameters

Visit official site
Category AI models

What Tencent-Hunyuan-Large is

The model is described as an open-source mixture-of-experts model built for natural language, vision and science tasks.

What you can do with it

  • Answer questions on long text
  • Summarise documents and generate content
  • Produce image descriptions

Model notes

  • Pretraining supports up to 256K
  • Instruct supports 128K sequences
  • Synthetic data aids generalisation

Pros & cons

✓ What we like

  • A vLLM-backend framework is provided
  • FP8 quantisation is supported

! What to watch out for

  • Large parameter counts are stated
  • Inference uses the provided framework

FAQ

How many parameters?

389 billion total and 52 billion active.

What context lengths?

256K for pretraining and 128K for Instruct.

Last reviewed: 2026-11-15

More LLM API platform tools

View all →

How we review