Tag
GGUF quantizations of DeepSeek V4 Flash in 2-bit, 3-bit, and 4-bit precisions, made available on Hugging Face for local inference with tools like llama.cpp and Ollama.