What are the hardware requirements for DeepSeek V4 Flash?

DeepSeek V4 Flash requires approximately 148.7 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 Flash and how large is it?

DeepSeek V4 Flash is developed by deepseek-ai, a model with 158.1b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does DeepSeek V4 Flash perform on standard benchmarks?

DeepSeek V4 Flash has the following published scores: LMSYS ELO 1436. Compare against reference models on this page or on the LLM Explorer leaderboards.

DeepSeek V4 Flash by deepseek-ai — VRAM 148.7GB, 1024K context

Name: DeepSeek V4 Flash
Author: deepseek-ai

DeepSeek V4 Flash is an open-source language model by deepseek-ai. Features: 158.1b LLM, VRAM: 148.7GB, Context: 1024K, License: mit, LLM Explorer Score: 0.62, ELO: 1436.

Arxiv:2606.19348 8-bit Conversational Deepseek v4 Deploy:azure Endpoints compatible Eval-results Fp8 Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash

DeepSeek V4 Flash Benchmarks

LMSys ELO 1436 vs 1272 (so35) 12.9%

^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 Flash Parameters and Internals

LLM Name	DeepSeek V4 Flash
Repository 🤗	https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash
Model Size	158.1b
Required VRAM	148.7 GB
Updated	2026-07-06
Maintainer	deepseek-ai
Model Type	deepseek_v4
Model Files	1.1 GB: 1-of-46 3.6 GB: 2-of-46 3.6 GB: 3-of-46 3.6 GB: 4-of-46 3.6 GB: 5-of-46 3.6 GB: 6-of-46 3.6 GB: 7-of-46 3.6 GB: 8-of-46 3.6 GB: 9-of-46 3.6 GB: 10-of-46 3.6 GB: 11-of-46 3.6 GB: 12-of-46 3.6 GB: 13-of-46 3.6 GB: 14-of-46 3.6 GB: 15-of-46 3.6 GB: 16-of-46 3.6 GB: 17-of-46 3.6 GB: 18-of-46 3.6 GB: 19-of-46 3.6 GB: 20-of-46 3.6 GB: 21-of-46 3.6 GB: 22-of-46 3.6 GB: 23-of-46 3.6 GB: 24-of-46 3.6 GB: 25-of-46 3.6 GB: 26-of-46 3.6 GB: 27-of-46 3.6 GB: 28-of-46 3.6 GB: 29-of-46 3.6 GB: 30-of-46 3.6 GB: 31-of-46 3.6 GB: 32-of-46 3.6 GB: 33-of-46 3.6 GB: 34-of-46 3.6 GB: 35-of-46 3.6 GB: 36-of-46 3.6 GB: 37-of-46 3.6 GB: 38-of-46 3.6 GB: 39-of-46 3.6 GB: 40-of-46 3.6 GB: 41-of-46 3.6 GB: 42-of-46
Model Architecture	DeepseekV4ForCausalLM
License	mit
Context Length	1048576
Model Max Length	1048576
Transformers Version	4.57.1
Vocabulary Size	129280
Torch Data Type	bfloat16

Rank the DeepSeek V4 Flash 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

Quantized Models of the DeepSeek V4 Flash

Model	Likes	Downloads	VRAM
DeepSeek V4 Flash GGUF	249	30807	0 GB
DeepSeek V4 Flash 2bit M DQ	2	785	106 GB
DeepSeek V4 Flash 3bit DQ	2	696	128 GB
...seek Ai DeepSeek V4 Flash 3bit	1	477	124 GB
...seek Ai DeepSeek V4 Flash 6bit	1	367	230 GB
...seek Ai DeepSeek V4 Flash 8bit	12	3577	215 GB
...seek Ai DeepSeek V4 Flash 4bit	5	1223	159 GB

Best Alternatives to DeepSeek V4 Flash

Best Alternatives	Context / RAM	Downloads	Likes
DeepSeek V4 Flash	1024K / 141.5 GB	1002	70