How much VRAM does gemma-4-26B-A4B-it-uncensored-heretic need?

Q4_K_M is exactly 16,796,011,744 bytes (15.64 GiB) in weights. Add the KV cache, which depends on context length, plus roughly half a gigabyte of runtime overhead.

How large is gemma-4-26B-A4B-it-uncensored-heretic's KV cache?

1.54 GiB at 32K context with an f16 cache, computed per layer from the model's architecture — most of its layers use sliding-window attention, so a flat formula overstates this substantially. Quantizing the cache to q8_0 roughly halves it.

Is gemma-4-26B-A4B-it-uncensored-heretic a mixture-of-experts model?

Yes — 128 experts with null 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.

llmfan46 · vision language · mixture of experts

gemma-4-26B-A4B-it-uncensored-heretic

llmfan46/gemma-4-26B-A4B-it-uncensored-heretic

gemma-4-26B-A4B-it-uncensored-heretic at Q4_K_M is exactly 16,796,011,744 bytes (15.64 GiB / 16.80 GB) — an effective 5.207 bits per weight, not the nominal 4. Its KV cache at 32K is 1.54 GiB, not the 7.50 GiB a flat formula predicts.

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

Parameters

25.8B

total, not active

Architecture

gemma4

30 layers

Context

262,144

native (config.json)

License

apache-2.0

Shipped quantizations

● exact bytes, summed from published files

Quant	Size●	Exact bytes●	Effective bpw●	Tensors●	Publisher
Q2_K	9.86 GiB	10,582,733,024	3.281	—	mradermacher
Q3_K_S	11.38 GiB	12,222,405,344	3.789	—	mradermacher
Q3_K_M	12.37 GiB	13,286,729,440	4.119	—	mradermacher
Q3_K_M	12.37 GiB	13,286,732,800	4.119	—	llmfan46
Q3_K_L	12.88 GiB	13,824,484,064	4.286	—	mradermacher
Q3_K_L	12.88 GiB	13,824,487,424	4.286	—	llmfan46
IQ4_XS	13.10 GiB	14,063,804,640	4.360	—	mradermacher
Q4_K_S	14.40 GiB	15,464,820,960	4.794	—	mradermacher
Q4_K_S	14.40 GiB	15,464,824,320	4.794	—	llmfan46
Q4_K_M	15.64 GiB	16,796,011,744	5.207	—	mradermacher
Q4_K_M	15.64 GiB	16,796,015,104	5.207	—	llmfan46
Q5_K_S	16.75 GiB	17,986,729,184	5.576	—	mradermacher
Q5_K_S	16.75 GiB	17,986,732,544	5.576	—	llmfan46
Q5_K_M	17.82 GiB	19,132,886,240	5.931	—	mradermacher
Q5_K_M	17.82 GiB	19,132,889,600	5.931	—	llmfan46
Q6_K	21.08 GiB	22,638,395,104	7.018	—	mradermacher
Q6_K	21.08 GiB	22,638,398,464	7.018	—	llmfan46
Q8_0	25.02 GiB	26,859,855,072	8.327	—	mradermacher
Q8_0	25.02 GiB	26,859,858,432	8.327	—	llmfan46
BF16	47.04 GiB	50,505,134,592	15.657	—	llmfan46

KV cache by context

computed per layer — this model uses sliding-window attention

Context	KV cache (f16)●	Flat formula	Overstated by	Full / windowed / recurrent
4,096	0.45 GiB	0.94 GiB	2.09×	5 / 25 / 0
8,192	0.61 GiB	1.88 GiB	3.10×	5 / 25 / 0
16,384	0.92 GiB	3.75 GiB	4.09×	5 / 25 / 0
32,768	1.54 GiB	7.50 GiB	4.86×	5 / 25 / 0
65,536	2.79 GiB	15.00 GiB	5.37×	5 / 25 / 0
131,072	5.29 GiB	30.00 GiB	5.67×	5 / 25 / 0

25 of 30 layers cache only a 1,024-token window rather than the full context, on a period of . Figures assume the default configuration; --swa-full disables the saving entirely.

Compare with

same modality, comparable size

vs Qwen3.6-35B-A3B vs Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTP vs Qwen3.5-9B vs gemma-4-26B-A4B-it vs gemma-4-12B-it vs Qwen3.5-4B All alternatives →Will it fit on my card? →

Why other calculators give a different number

A parameters × bits ÷ 8 estimate puts Q4_K_M at roughly 13.52 GiB. The real file is 15.64 GiB, because a quantization is a mixture and some tensors are always kept at higher precision. The larger discrepancy is the cache: a flat formula gives 7.50 GiB at 32K context where the real figure is 1.54 GiB, because most of this model's layers cache a fixed window rather than the whole context.

Architecture

from config.json

Layers

Attention heads

KV heads

Head dim

256

Hidden size

2816

Vocab

262,144

Sliding window

1024

SWA period

—

MLA

Experts

128

Experts per token

—

use_sliding_window

—

Questions people ask

How much VRAM does gemma-4-26B-A4B-it-uncensored-heretic need?: Q4_K_M is exactly 16,796,011,744 bytes (15.64 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
How large is gemma-4-26B-A4B-it-uncensored-heretic's KV cache?: 1.54 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.
Is gemma-4-26B-A4B-it-uncensored-heretic a mixture-of-experts model?: Yes — 128 experts, null 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 gemma-4-26B-A4B-it-uncensored-heretic 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.