What are the hardware requirements for DeepSeek V4 Pro 4bit?

DeepSeek V4 Pro 4bit requires approximately 206.2 GB of VRAM and supports a context window of 1024K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed DeepSeek V4 Pro 4bit and how large is it?

DeepSeek V4 Pro 4bit is developed by mlx-community, a model with 1573b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does DeepSeek V4 Pro 4bit perform on standard benchmarks?

DeepSeek V4 Pro 4bit has the following published scores: LMSYS ELO 1455. Compare against reference models on this page or on the LLM Explorer leaderboards.

DeepSeek V4 Pro 4bit by mlx-community — VRAM 206.2GB, 1024K context

Name: DeepSeek V4 Pro 4bit
Author: mlx-community

DeepSeek V4 Pro 4bit is an open-source language model by mlx-community. Features: 1573b LLM, VRAM: 206.2GB, Context: 1024K, Quantized, LLM Explorer Score: 0.41, ELO: 1455.

4-bit 4bit Conversational Deepseek v4 En Mlx Quantized Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/mlx-community/DeepSeek-V4-Pro-4bit

DeepSeek V4 Pro 4bit Benchmarks

LMSys ELO 1455 vs 1272 (so35) 14.4%

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

DeepSeek V4 Pro 4bit Parameters and Internals

LLM Name	DeepSeek V4 Pro 4bit
Repository 🤗	https://huggingface.co/mlx-community/DeepSeek-V4-Pro-4bit
Model Size	1573b
Required VRAM	206.2 GB
Updated	2026-07-25
Maintainer	mlx-community
Model Type	deepseek_v4
Model Files	5.2 GB: 1-of-183 4.5 GB: 2-of-183 4.7 GB: 3-of-183 4.5 GB: 4-of-183 4.5 GB: 5-of-183 4.7 GB: 6-of-183 4.5 GB: 7-of-183 4.5 GB: 8-of-183 4.7 GB: 9-of-183 4.5 GB: 10-of-183 4.5 GB: 11-of-183 4.7 GB: 12-of-183 4.5 GB: 13-of-183 4.5 GB: 14-of-183 4.7 GB: 15-of-183 4.5 GB: 16-of-183 4.5 GB: 17-of-183 4.7 GB: 18-of-183 4.5 GB: 19-of-183 4.5 GB: 20-of-183 4.7 GB: 21-of-183 4.5 GB: 22-of-183 4.5 GB: 23-of-183 4.7 GB: 24-of-183 4.5 GB: 25-of-183 4.5 GB: 26-of-183 4.7 GB: 27-of-183 4.5 GB: 28-of-183 4.5 GB: 29-of-183 4.7 GB: 30-of-183 4.5 GB: 31-of-183 4.5 GB: 32-of-183 4.7 GB: 33-of-183 4.5 GB: 34-of-183 4.5 GB: 35-of-183 4.7 GB: 36-of-183 4.5 GB: 37-of-183 4.5 GB: 38-of-183 4.7 GB: 39-of-183 4.5 GB: 40-of-183 4.5 GB: 41-of-183 4.7 GB: 42-of-183 4.5 GB: 43-of-183 4.5 GB: 44-of-183 4.7 GB: 45-of-183
Supported Languages	en
Quantization Type	4bit
Model Architecture	DeepseekV4ForCausalLM
Context Length	1048576
Model Max Length	1048576
Transformers Version	4.57.1
Vocabulary Size	129280
Torch Data Type	bfloat16

Rank the DeepSeek V4 Pro 4bit capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to DeepSeek V4 Pro 4bit

Best Alternatives	Context / RAM	Downloads	Likes
DeepSeek V4 Pro 8bit	1024K / 204.9 GB	1369	2
DeepSeek V4 Pro	1024K / 209.8 GB	4940	2

Note: green Score (e.g. "73.2") means that the model is better than mlx-community/DeepSeek-V4-Pro-4bit.