LLM EXPLORER 62,155 MODELS INDEXED

DeepSeek V4 Pro Bf16 by mlx-community

By mlx-community · 401 downloads
HF Repo ↗ Copy URL

DeepSeek V4 Pro Bf16 is an open-source language model by mlx-community. Features: 1573b LLM, VRAM: 211.1GB, Context: 1024K, LLM Explorer Score: 0.25.

  4-bit   Conversational   Deepseek v4   En   Mlx   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/mlx-community/DeepSeek-V4-Pro-bf16 

DeepSeek V4 Pro Bf16 Parameters and Internals

LLM NameDeepSeek V4 Pro Bf16
Repository 🤗https://huggingface.co/mlx-community/DeepSeek-V4-Pro-bf16 
Model Size1573b
Required VRAM211.1 GB
Updated2026-09-25
Maintainermlx-community
Model Typedeepseek_v4
Model Files  2.5 GB: 1-of-185   4.5 GB: 2-of-185   4.5 GB: 3-of-185   5.3 GB: 4-of-185   4.5 GB: 5-of-185   4.5 GB: 6-of-185   5.3 GB: 7-of-185   4.5 GB: 8-of-185   4.5 GB: 9-of-185   5.2 GB: 10-of-185   4.5 GB: 11-of-185   4.5 GB: 12-of-185   5.3 GB: 13-of-185   4.5 GB: 14-of-185   4.5 GB: 15-of-185   5.2 GB: 16-of-185   4.5 GB: 17-of-185   4.5 GB: 18-of-185   5.3 GB: 19-of-185   4.5 GB: 20-of-185   4.5 GB: 21-of-185   5.2 GB: 22-of-185   4.5 GB: 23-of-185   4.5 GB: 24-of-185   5.3 GB: 25-of-185   4.5 GB: 26-of-185   4.5 GB: 27-of-185   5.2 GB: 28-of-185   4.5 GB: 29-of-185   4.5 GB: 30-of-185   5.3 GB: 31-of-185   4.5 GB: 32-of-185   4.5 GB: 33-of-185   5.2 GB: 34-of-185   4.5 GB: 35-of-185   4.5 GB: 36-of-185   5.3 GB: 37-of-185   4.5 GB: 38-of-185   4.5 GB: 39-of-185   5.2 GB: 40-of-185   4.5 GB: 41-of-185   4.5 GB: 42-of-185   5.3 GB: 43-of-185   4.5 GB: 44-of-185   4.5 GB: 45-of-185
Supported Languagesen
Model ArchitectureDeepseekV4ForCausalLM
Context Length1048576
Model Max Length1048576
Transformers Version4.57.1
Vocabulary Size129280
Torch Data Typebfloat16

Best Alternatives to DeepSeek V4 Pro Bf16

Best Alternatives
Context / RAM
Downloads
Likes
DeepSeek V4 Pro1024K / 209.8 GB49402
DeepSeek V4 Pro 4bit1024K / 206.2 GB15912
DeepSeek V4 Pro 8bit1024K / 204.9 GB13692
Note: green Score (e.g. "73.2") means that the model is better than mlx-community/DeepSeek-V4-Pro-bf16.