How much VRAM does DeepSeek-V4-Flash need?

UD-IQ1_S is exactly 82,539,237,024 bytes (76.87 GiB) in weights. Add the KV cache, which depends on context length, plus roughly half a gigabyte of runtime overhead.

Is DeepSeek-V4-Flash a mixture-of-experts model?

Yes — 256 experts with 6 routed per token. Every expert must be resident in memory, but only the routed ones are read per token, so its memory requirement and its speed behave very differently.

deepseek-ai · text · mixture of experts

DeepSeek-V4-Flash

deepseek-ai/DeepSeek-V4-Flash

DeepSeek-V4-Flash at UD-IQ1_S is exactly 82,539,237,024 bytes (76.87 GiB / 82.54 GB) — an effective 4.177 bits per weight, not the nominal 1.

From the file· summed from 3 file(s)From the file· KV per layer

Parameters

158B

total, not active

Architecture

deepseek4-dspark

43 layers

Context

1,048,576

native (config.json)

License

mit

Shipped quantizations

● exact bytes, summed from published files

Quant	Size●	Exact bytes●	Effective bpw●	Tensors●	Publisher
UD-IQ1_S3 shards	76.87 GiB	82,539,237,024	4.177	—	unsloth
UD-IQ1_M3 shards	80.93 GiB	86,901,313,152	4.398	—	unsloth
UD-IQ2_XXS3 shards	84.62 GiB	90,860,736,128	4.598	—	unsloth
UD-IQ2_M3 shards	84.68 GiB	90,926,927,488	4.602	—	unsloth
UD-IQ3_XXS4 shards	95.93 GiB	102,999,887,616	5.213	—	unsloth
UD-IQ3_S4 shards	109.25 GiB	117,310,852,864	5.937	—	unsloth
UD-Q3_K_M4 shards	120.44 GiB	129,319,997,216	6.545	—	unsloth
UD-IQ4_XS4 shards	128.43 GiB	137,903,959,808	6.979	—	unsloth
UD-IQ4_NL4 shards	128.43 GiB	137,903,959,808	6.979	—	unsloth
MXFP44 shards	145.29 GiB	156,004,658,464	7.896	—	bartowski

KV cache by context

unresolved

This model declares a 128-token sliding window, but we could not establish which layers use it. Its architecture publishes the layout as a per-layer array inside the model file rather than as a period in config.json, and we have not yet ingested that array.

A flat context × layers × heads figure would be substantially too high, so we are not showing one. This is tracked as a known gap rather than filled with a guess.

Compare with

same modality, comparable size

vs Qwen3-Coder-30B-A3B-Instruct vs Hy3 vs Qwen-AgentWorld-35B-A3B vs GLM-5.2 vs Ornith-1.0-35B vs Qwen3-Coder-Next All alternatives →Will it fit on my card? →

Why other calculators give a different number

A parameters × bits ÷ 8 estimate puts UD-IQ1_S at roughly 82.81 GiB. The real file is 76.87 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

from config.json

Layers

Attention heads

KV heads

Head dim

512

Hidden size

4096

Vocab

129,280

Sliding window

128

SWA period

—

MLA

Experts

256

Experts per token

use_sliding_window

—

Questions people ask

How much VRAM does DeepSeek-V4-Flash need?: UD-IQ1_S is exactly 82,539,237,024 bytes (76.87 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
Is DeepSeek-V4-Flash a mixture-of-experts model?: Yes — 256 experts, 6 routed per token. Every expert must be resident, but only the routed ones are read per token, which is why its memory requirement and its speed behave very differently.
Which quantization of DeepSeek-V4-Flash should I use?: Q4_K_M is the usual default. Pick the largest quantization that fits your card at the context you actually need — the table above gives exact sizes for every one published.