How much VRAM does ZAYA1-8B-Coder need?

Q4_K_M is exactly 5,567,581,024 bytes (5.19 GiB) in weights. Add the KV cache, which depends on context length, plus roughly half a gigabyte of runtime overhead.

How large is ZAYA1-8B-Coder's KV cache?

2.50 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.

Is ZAYA1-8B-Coder a mixture-of-experts model?

Yes — 16 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.

josephmayo · text · mixture of experts

ZAYA1-8B-Coder

josephmayo/ZAYA1-8B-Coder

ZAYA1-8B-Coder at Q4_K_M is exactly 5,567,581,024 bytes (5.19 GiB / 5.57 GB) — an effective 5.038 bits per weight, not the nominal 4. Its KV cache at 32K is 2.50 GiB.

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

Parameters

8.8B

total, not active

Architecture

zaya

80 layers

Context

131,072

native (config.json)

License

—

Shipped quantizations

● exact bytes, summed from published files

Quant	Size●	Exact bytes●	Effective bpw●	Tensors●	Publisher
Q4_K_M	5.19 GiB	5,567,581,024	5.038	—	josephmayo
Q6_K	6.85 GiB	7,353,195,104	6.654	—	josephmayo
Q8_0	8.83 GiB	9,485,673,760	8.584	—	josephmayo

KV cache by context

computed per layer

Context	KV cache (f16)●	Flat formula	Overstated by	Full / windowed / recurrent
4,096	0.31 GiB	0.31 GiB	—	80 / 0 / 0
8,192	0.63 GiB	0.63 GiB	—	80 / 0 / 0
16,384	1.25 GiB	1.25 GiB	—	80 / 0 / 0
32,768	2.50 GiB	2.50 GiB	—	80 / 0 / 0
65,536	5.00 GiB	5.00 GiB	—	80 / 0 / 0
131,072	10.00 GiB	10.00 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-4-12B-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 4.63 GiB. The real file is 5.19 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

2048

Vocab

262,272

Sliding window

none

SWA period

—

MLA

Experts

Experts per token

—

use_sliding_window

—

Questions people ask

How much VRAM does ZAYA1-8B-Coder need?: Q4_K_M is exactly 5,567,581,024 bytes (5.19 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
How large is ZAYA1-8B-Coder's KV cache?: 2.50 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 ZAYA1-8B-Coder a mixture-of-experts model?: Yes — 16 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 ZAYA1-8B-Coder 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.