Skip to content

DeepSeek V4 Flash VRAM requirements

How much VRAM DeepSeek V4 Flash needs
FP4 + FP8 weights: 155 GBWith 8K context: 156 GBWith 32K context: 156 GB
Parameters
284B
Active per token
13B
Longest context
1M
Released
Jul 2026
License
MIT
Strongest at
Docs

How much VRAM DeepSeek V4 Flash needs

Memory DeepSeek V4 Flash needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 128K
As released (FP4 + FP8)155 GB156 GB156 GB157 GB
3-bit129 GB130 GB130 GB131 GB
2-bit99.2 GB100 GB100 GB101 GB
1.6-bit54.6 GB55.4 GB55.6 GB56.1 GB

One conversation, 16-bit context cache, runtime overhead included.

GPUs and Macs that run DeepSeek V4 Flash

The best way DeepSeek V4 Flash runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 4060Too bigPDF: DeepSeek V4 Flash on RTX 4060 report
RTX 5060Too bigPDF: DeepSeek V4 Flash on RTX 5060 report
RTX 3070Too bigPDF: DeepSeek V4 Flash on RTX 3070 report
RTX 3080 10 GBToo bigPDF: DeepSeek V4 Flash on RTX 3080 10 GB report
RTX 3060 12 GBToo bigPDF: DeepSeek V4 Flash on RTX 3060 12 GB report
Arc B580Too bigPDF: DeepSeek V4 Flash on Arc B580 report
RTX 4070Too bigPDF: DeepSeek V4 Flash on RTX 4070 report
RTX 4070 SuperToo bigPDF: DeepSeek V4 Flash on RTX 4070 Super report
RTX 5070Too bigPDF: DeepSeek V4 Flash on RTX 5070 report
RTX 4060 Ti 16 GBToo bigPDF: DeepSeek V4 Flash on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GBToo bigPDF: DeepSeek V4 Flash on RTX 5060 Ti 16 GB report
Arc A770 16 GBToo bigPDF: DeepSeek V4 Flash on Arc A770 16 GB report
Radeon RX 9070 XTToo bigPDF: DeepSeek V4 Flash on Radeon RX 9070 XT report
RTX 4070 Ti SuperToo bigPDF: DeepSeek V4 Flash on RTX 4070 Ti Super report
RTX 4080 SuperToo bigPDF: DeepSeek V4 Flash on RTX 4080 Super report
RTX 5070 TiToo bigPDF: DeepSeek V4 Flash on RTX 5070 Ti report
RTX 5080Too bigPDF: DeepSeek V4 Flash on RTX 5080 report
Radeon RX 7900 XTXToo bigPDF: DeepSeek V4 Flash on Radeon RX 7900 XTX report
RTX 3090Too bigPDF: DeepSeek V4 Flash on RTX 3090 report
RTX 4090Too bigPDF: DeepSeek V4 Flash on RTX 4090 report
Radeon AI PRO R9700Too bigPDF: DeepSeek V4 Flash on Radeon AI PRO R9700 report
RTX 5090Too bigPDF: DeepSeek V4 Flash on RTX 5090 report
2× RTX 3090Too bigPDF: DeepSeek V4 Flash on 2× RTX 3090 report
2× RTX 4090Too bigPDF: DeepSeek V4 Flash on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GBToo bigPDF: DeepSeek V4 Flash on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 AdaToo bigPDF: DeepSeek V4 Flash on RTX 6000 Ada report
L40SToo bigPDF: DeepSeek V4 Flash on L40S report
2× RTX 5090Too bigPDF: DeepSeek V4 Flash on 2× RTX 5090 report
4× RTX 3090Too bigPDF: DeepSeek V4 Flash on 4× RTX 3090 report
RTX PRO 6000 BlackwellToo bigPDF: DeepSeek V4 Flash on RTX PRO 6000 Blackwell report
Mac mini M4 16 GBToo bigPDF: DeepSeek V4 Flash on Mac mini M4 16 GB report
Mac mini M4 24 GBToo bigPDF: DeepSeek V4 Flash on Mac mini M4 24 GB report
Mac mini M4 32 GBToo bigPDF: DeepSeek V4 Flash on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GBToo bigPDF: DeepSeek V4 Flash on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GBToo bigPDF: DeepSeek V4 Flash on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GBToo bigPDF: DeepSeek V4 Flash on Mac mini M4 Pro 64 GB report
M4 Max 128 GBToo bigPDF: DeepSeek V4 Flash on M4 Max 128 GB report
M5 Max 128 GBToo bigPDF: DeepSeek V4 Flash on M5 Max 128 GB report
M3 Ultra 96 GBToo bigPDF: DeepSeek V4 Flash on M3 Ultra 96 GB report
M3 Ultra 256 GBFP4 + FP81M24 tok/s7.7 sPDF: DeepSeek V4 Flash on M3 Ultra 256 GB report
M3 Ultra 512 GBFP4 + FP81M24 tok/s7.7 sPDF: DeepSeek V4 Flash on M3 Ultra 512 GB report
M5 Ultra 512 GBFP4 + FP81M33 tok/s2.4 sPDF: DeepSeek V4 Flash on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GBToo bigPDF: DeepSeek V4 Flash on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GBToo bigPDF: DeepSeek V4 Flash on Ryzen AI Max+ 395 128 GB report
RTX Spark2-bit1M28 tok/s2.8 sPDF: DeepSeek V4 Flash on RTX Spark report
DGX Spark2-bit1M25 tok/s2.8 sPDF: DeepSeek V4 Flash on DGX Spark report
Dual DGX SparkFP4 + FP81M9.9 tok/s2.0 sPDF: DeepSeek V4 Flash on Dual DGX Spark report
DGX Station GB300FP4 + FP81M157 tok/s0.2 sPDF: DeepSeek V4 Flash on DGX Station GB300 report
2× RTX PRO 6000 BlackwellFP4 + FP81M57 tok/s0.4 sPDF: DeepSeek V4 Flash on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 BlackwellFP4 + FP81M57 tok/s0.2 sPDF: DeepSeek V4 Flash on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 BlackwellFP4 + FP81M57 tok/s0.1 sPDF: DeepSeek V4 Flash on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXMToo bigPDF: DeepSeek V4 Flash on A100 80 GB SXM report
H100 SXMToo bigPDF: DeepSeek V4 Flash on H100 SXM report
4× H200 NVLFP4 + FP81M122 tok/s0.2 sPDF: DeepSeek V4 Flash on 4× H200 NVL report
8× H200 NVLFP4 + FP81M122 tok/s0.1 sPDF: DeepSeek V4 Flash on 8× H200 NVL report

Best fit is the highest quality, up to 8-bit, that leaves 16K of context. First reply is the wait for a 2K prompt. Speeds are rough, plus or minus 25%.

How to run it

llama.cpp
llama-server -hf unsloth/DeepSeek-V4-Flash-0731-GGUF
vLLM
vllm serve deepseek-ai/DeepSeek-V4-Flash-0731
MLX (Mac)
mlx_lm.chat --model mlx-community/DeepSeek-V4-Flash-0731-2.4bit-mixed

Questions

How much VRAM does DeepSeek V4 Flash need?

DeepSeek V4 Flash needs 155 GB for its weights as released (FP4 + FP8), 156 GB with an 8K context and 156 GB with 32K.

What is the minimum VRAM for DeepSeek V4 Flash?

About 100 GB at 2-bit with an 8K context, the lowest precision we recommend for a model this size. The smallest machine here that runs it without spilling is the M3 Ultra 256 GB.

How big is DeepSeek V4 Flash?

284B parameters (13B active per token): 155 GB of weights as released (FP4 + FP8).

Can DeepSeek V4 Flash run on an RTX 4090?

Not on one card. The smallest machine here that runs it is the M3 Ultra 256 GB.

More models