DeepSeek/Mixture of Experts

DeepSeek V4 Flash

Name: DeepSeek V4 Flash
Author: DeepSeek

chatcodingreasoningmultilingualmathagentictool_use

284B

Parameters (13B active)

1024K

Context length

Benchmarks

Quantizations

97K

HF downloads

Architecture

MoE

Released

2026-04-24

Layers

KV Heads

Head Dim

512

Family

deepseek

Quantization Options

Quant	Bits	VRAM	Quality
IQ2_XXS	2.38	85.0 GB	low
IQ2_M	2.93	104.5 GB	low
Q2_K	3.16	112.7 GB	low
IQ3_XXS	3.25	115.9 GB	low
IQ3_XS	3.5	124.7 GB	low
Q3_K_S	3.64	129.7 GB	low
IQ3_M	3.76	134.0 GB	low
Q3_K_M	4	142.5 GB	low
Q3_K_L	4.3	153.1 GB	moderate
IQ4_XS	4.46	158.8 GB	moderate
Q4_K_S	4.67	166.3 GB	moderate
Q4_K_M	4.89	174.1 GB	good
Q5_K_S	5.57	198.2 GB	good
Q5_K_M	5.7	202.8 GB	good
Q6_K	6.56	233.4 GB	excellent
Q8_0	8.5	302.2 GB	lossless
FP16	16	568.5 GB	lossless

Select your GPU above to see speed estimates and compatibility for each quantization.

▸ READY TO RUN THIS?RENT BY THE HOUR

RENT A GPU AND RUN DEEPSEEK V4 FLASH NOW

Rent on RunPod →Or Vast.ai →

Spin up an A100 / H100 / 4090 in ~60s. Pay by the second. Cancel anytime.

Community Ratings

Loading ratings...

Benchmarks (7)

LiveCodeBench91.6

GPQA Diamond88.1

MMLU-PRO86.2

SWE-bench79.0

HumanEval69.5

Terminal-Bench56.9

HLE34.8

Run this model

▸Easiest way to get started·Beginners

DOCS ↗

curl -fsSL https://ollama.com/install.sh | sh

$ollama run deepseek:284b-q4_K_M

Tag may need adjustment — check ollama.com/library/deepseek for available tags.

▸ SETUP GUIDE

Auto-setup with fitmyllm CLI

Detects your GPU, recommends the best model, downloads it, and starts chatting — zero config. Benchmarks your speed and contributes anonymous data to improve predictions.

pip install fitmyllmthen run fitmyllmLearn more

Auto-detect GPULive tok/s in chatSpeed benchmarks9 inference engines

HuggingFace GGUF Downloads Build Hardware