How much VRAM does Tinman-gemma4-companion-merged need?

Q4_K_M is exactly 5,302,272,544 bytes (4.94 GiB) in weights. Add the KV cache, which depends on context length, plus roughly half a gigabyte of runtime overhead.

How large is Tinman-gemma4-companion-merged's KV cache?

0.51 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.

Tinman-Lab · text

Tinman-gemma4-companion-merged

Tinman-Lab/Tinman-gemma4-companion-merged

Tinman-gemma4-companion-merged at Q4_K_M is exactly 5,302,272,544 bytes (4.94 GiB / 5.30 GB) — an effective 5.342 bits per weight, not the nominal 4. Its KV cache at 32K is 0.51 GiB, not the 2.63 GiB a flat formula predicts.

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

Parameters

7.9B

Architecture

gemma4

42 layers

Context

131,072

native (config.json)

License

apache-2.0

Shipped quantizations

● exact bytes, summed from published files

Quant	Size●	Exact bytes●	Effective bpw●	Tensors●	Publisher
Q4_K_M	4.94 GiB	5,302,272,544	5.342	—	Tinman-Lab
Q8_0	7.43 GiB	7,972,725,280	8.032	—	Tinman-Lab
BF16	13.92 GiB	14,942,970,400	15.054	—	Tinman-Lab

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	0.12 GiB	0.33 GiB	2.67×	7 / 35 / 0
8,192	0.18 GiB	0.66 GiB	3.69×	7 / 35 / 0
16,384	0.29 GiB	1.31 GiB	4.57×	7 / 35 / 0
32,768	0.51 GiB	2.63 GiB	5.19×	7 / 35 / 0
65,536	0.94 GiB	5.25 GiB	5.57×	7 / 35 / 0
131,072	1.82 GiB	10.50 GiB	5.77×	7 / 35 / 0

35 of 42 layers cache only a 512-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-Coder-30B-A3B-Instruct vs Qwen3.6-27B vs Qwen3.8-27B vs gemma-4-12B-it-qat-q4_0-unquantized vs gemma-4-E4B-it vs Qwen3-30B-A3B-Thinking-2507 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 4.16 GiB. The real file is 4.94 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 2.63 GiB at 32K context where the real figure is 0.51 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

2560

Vocab

262,144

Sliding window

512

SWA period

—

MLA

Experts

—

Experts per token

—

use_sliding_window

—

Questions people ask

How much VRAM does Tinman-gemma4-companion-merged need?: Q4_K_M is exactly 5,302,272,544 bytes (4.94 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
How large is Tinman-gemma4-companion-merged's KV cache?: 0.51 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 Tinman-gemma4-companion-merged 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.