---
base_model:
- deepseek-ai/DeepSeek-V3
---

UPDATE Jan 4th 2025: Support for DeepSeek-V3 has been merged, you can now pull from the master branch of llama.cpp. The versions uploaded in this repo are already requanted to support the changes in the tensor names


Initial preview for the GGUF quantized version of [deepseek-ai/DeepSeek-V3](https://huggingface.co/deepseek-ai/DeepSeek-V3)

It needs this PR commit to work: https://github.com/ggerganov/llama.cpp/pull/11049

Thanks to Fairydreaming for the PR! 

Note: no multi-token prediction (MTP) support