LLM EXPLORER 58,425 MODELS INDEXED

DeepSeek V4 Flash 0731 W4A16 AutoRound by Intel

By Intel · 179 downloads
HF Repo ↗ Copy URL

DeepSeek V4 Flash 0731 W4A16 AutoRound is an open-source language model by Intel. Features: 43.4b LLM, VRAM: 148.1GB, Context: 1024K, License: mit, LLM Explorer Score: 0.32.

  Arxiv:2309.05516   4-bit   Auto-round   Deepseek v4   Endpoints compatible   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/Intel/DeepSeek-V4-Flash-0731-W4A16-AutoRound 

DeepSeek V4 Flash 0731 W4A16 AutoRound Parameters and Internals

LLM NameDeepSeek V4 Flash 0731 W4A16 AutoRound
Repository 🤗https://huggingface.co/Intel/DeepSeek-V4-Flash-0731-W4A16-AutoRound 
Model Size43.4b
Required VRAM148.1 GB
Updated2026-08-14
MaintainerIntel
Model Typedeepseek_v4
Model Files  1.1 GB: 1-of-48   3.5 GB: 2-of-48   3.5 GB: 3-of-48   3.5 GB: 4-of-48   3.5 GB: 5-of-48   3.5 GB: 6-of-48   3.5 GB: 7-of-48   3.5 GB: 8-of-48   3.5 GB: 9-of-48   3.5 GB: 10-of-48   3.5 GB: 11-of-48   3.5 GB: 12-of-48   3.5 GB: 13-of-48   3.5 GB: 14-of-48   3.5 GB: 15-of-48   3.5 GB: 16-of-48   3.5 GB: 17-of-48   3.5 GB: 18-of-48   3.5 GB: 19-of-48   3.5 GB: 20-of-48   3.5 GB: 21-of-48   3.5 GB: 22-of-48   3.5 GB: 23-of-48   3.5 GB: 24-of-48   3.5 GB: 25-of-48   3.5 GB: 26-of-48   3.5 GB: 27-of-48   3.5 GB: 28-of-48   3.5 GB: 29-of-48   3.5 GB: 30-of-48   3.5 GB: 31-of-48   3.5 GB: 32-of-48   3.5 GB: 33-of-48   3.5 GB: 34-of-48   3.5 GB: 35-of-48   3.5 GB: 36-of-48   3.5 GB: 37-of-48   3.5 GB: 38-of-48   3.5 GB: 39-of-48   3.5 GB: 40-of-48   3.5 GB: 41-of-48   3.5 GB: 42-of-48   3.5 GB: 43-of-48
Model ArchitectureDeepseekV4ForCausalLM
Licensemit
Context Length1048576
Model Max Length1048576
Transformers Version4.57.1
Vocabulary Size129280
Torch Data Typebfloat16