How much VRAM does Kimi-Dev-72B need?

UD-IQ1_S is exactly 23,048,185,344 bytes (21.47 GiB) in weights. Add the KV cache, which depends on context length, plus roughly half a gigabyte of runtime overhead.

How large is Kimi-Dev-72B's KV cache?

10.00 GiB at 32K context with an f16 cache, computed per layer from the model's architecture. Quantizing the cache to q8_0 roughly halves it.

moonshotai · text

Kimi-Dev-72B

moonshotai/Kimi-Dev-72B

Kimi-Dev-72B at UD-IQ1_S is exactly 23,048,185,344 bytes (21.47 GiB / 23.05 GB) — an effective 2.536 bits per weight, not the nominal 1. Its KV cache at 32K is 10.00 GiB.

From the file· summed from 1 file(s)From the file· KV per layer

Parameters

72.7B

Architecture

qwen2

80 layers

Context

131,072

native (config.json)

License

mit

Shipped quantizations

● exact bytes, summed from published files

Quant	Size●	Exact bytes●	Effective bpw●	Tensors●	Publisher
UD-IQ1_S	21.47 GiB	23,048,185,344	2.536	—	unsloth
UD-IQ1_M	22.35 GiB	23,994,361,344	2.640	—	unsloth
UD-IQ2_XXS	23.94 GiB	25,700,427,264	2.828	—	unsloth
UD-IQ2_M	27.56 GiB	29,596,542,464	3.257	—	unsloth
UD-IQ3_XXS	29.67 GiB	31,856,879,104	3.505	—	unsloth
Q3_K_S	32.12 GiB	34,487,789,056	3.795	—	unsloth
Q3_K_M	35.11 GiB	37,698,725,376	4.148	—	unsloth
IQ4_XS	37.02 GiB	39,748,003,328	4.373	—	unsloth
IQ4_NL	38.48 GiB	41,315,624,448	4.546	—	unsloth
Q4_0	38.54 GiB	41,383,126,528	4.553	—	unsloth
Q4_1	42.56 GiB	45,697,885,696	5.028	—	unsloth
Q6_K2 shards	59.93 GiB	64,347,629,248	7.080	—	unsloth
Q8_02 shards	71.96 GiB	77,262,612,192	8.501	—	unsloth
BF163 shards	135.44 GiB	145,422,673,472	16.001	—	unsloth

KV cache by context

computed per layer

Context	KV cache (f16)●	Flat formula	Overstated by	Full / windowed / recurrent
4,096	1.25 GiB	1.25 GiB	—	80 / 0 / 0
8,192	2.50 GiB	2.50 GiB	—	80 / 0 / 0
16,384	5.00 GiB	5.00 GiB	—	80 / 0 / 0
32,768	10.00 GiB	10.00 GiB	—	80 / 0 / 0
65,536	20.00 GiB	20.00 GiB	—	80 / 0 / 0
131,072	40.00 GiB	40.00 GiB	—	80 / 0 / 0

Compare with

same modality, comparable size

vs Qwen3-Coder-30B-A3B-Instruct vs Qwen3.6-27B vs Qwen3.8-27B vs gemma-4-E4B-it vs Qwen3-30B-A3B-Thinking-2507 vs Qwen3-8B 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 38.09 GiB. The real file is 21.47 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

128

Hidden size

8192

Vocab

152,064

Sliding window

131072

SWA period

—

MLA

Experts

—

Experts per token

—

use_sliding_window

false

This model declares a sliding window but sets use_sliding_window: false, so the window is not applied. Honouring the field without the flag understates KV for the whole family.

Questions people ask

How much VRAM does Kimi-Dev-72B need?: UD-IQ1_S is exactly 23,048,185,344 bytes (21.47 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
How large is Kimi-Dev-72B's KV cache?: 10.00 GiB at 32K context with an f16 cache, computed per layer. Quantizing the cache to q8_0 roughly halves it, which is often the difference between a context length fitting and not.
Which quantization of Kimi-Dev-72B 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.