DeepSeek-V4-Pro#

  • Context Length: 163840

  • Model Name: DeepSeek-V4-Pro

  • Languages: en, zh

  • Abilities: chat, reasoning, hybrid, tools

  • Description: We present a preview version of DeepSeek-V4 series, including two strong Mixture-of-Experts (MoE) language models — DeepSeek-V4-Pro with 1.6T parameters (49B activated) and DeepSeek-V4-Flash with 284B parameters (13B activated) — both supporting a context length of one million tokens.

Specifications#

Model Spec 1 (pytorch, 1600 Billion)#

  • Model Format: pytorch

  • Model Size (in billions): 1600

  • Quantizations: none

  • Engines: Transformers

  • Model ID: deepseek-ai/DeepSeek-V4-Pro

  • Model Hubs: Hugging Face, ModelScope

Execute the following command to launch the model, remember to replace ${quantization} with your chosen quantization method from the options listed above:

xinference launch --model-engine ${engine} --model-name DeepSeek-V4-Pro --size-in-billions 1600 --model-format pytorch --quantization ${quantization}

Model Spec 2 (mlx, 1600 Billion)#

  • Model Format: mlx

  • Model Size (in billions): 1600

  • Quantizations: 4bit, 8bit, bf16

  • Engines: MLX

  • Model ID: mlx-community/DeepSeek-V4-Pro-{quantization}

  • Model Hubs: Hugging Face, ModelScope

Execute the following command to launch the model, remember to replace ${quantization} with your chosen quantization method from the options listed above:

xinference launch --model-engine ${engine} --model-name DeepSeek-V4-Pro --size-in-billions 1600 --model-format mlx --quantization ${quantization}