Hunyuan LLM
A self-hosted LLM with fast and slow reasoning modes, a 256K context window, and support for Transformers and vLLM.
Hunyuan-A13B is Tencent's language model for developers and researchers building self-hosted AI applications. It handles general text and reasoning tasks, with optimization for AI agent workloads.
- Reasoning modes: Applications can choose fast thinking or slower, more deliberate reasoning according to the task; slow thinking is the default.
- Long context: A 256K context window supports work with long documents and substantial task context.
- Agent workloads: The model is optimized for agent tasks, with evaluations on BFCL-v3, τ-Bench and C3-Bench.
- Quantized models: Tencent provides a downloadable FP8 Instruct model, produces INT4 models, and supports INT8, INT4 and FP8 deployment through vLLM.
Hunyuan-A13B works with Transformers and can run on your own infrastructure through vLLM. Its Mixture-of-Experts architecture has 80 billion total parameters and 13 billion active parameters, a distinction worth considering when comparing it with smaller models.
The repository uses a custom license.
Logotyper
