How much VRAM does next-1b need?

Q4_K_M is exactly 806,057,664 bytes (0.75 GiB) in weights. Add the KV cache, which depends on context length, plus roughly half a gigabyte of runtime overhead.

How large is next-1b's KV cache?

0.15 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.

thelamapi · text

next-1b

thelamapi/next-1b

next-1b at Q4_K_M is exactly 806,057,664 bytes (0.75 GiB / 0.81 GB) — an effective 6.449 bits per weight, not the nominal 4. Its KV cache at 32K is 0.15 GiB, not the 0.81 GiB a flat formula predicts.

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

Parameters

1000M

Architecture

gemma3

26 layers

Context

32,768

native (config.json)

License

mit

Shipped quantizations

● exact bytes, summed from published files

Quant	Size●	Exact bytes●	Effective bpw●	Tensors●	Publisher
Q4_K_M	0.75 GiB	806,057,664	6.449	—	mattritchey

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	0.04 GiB	0.10 GiB	2.74×	4 / 22 / 0
8,192	0.05 GiB	0.20 GiB	3.85×	4 / 22 / 0
16,384	0.08 GiB	0.41 GiB	4.84×	4 / 22 / 0
32,768	0.15 GiB	0.81 GiB	5.55×	4 / 22 / 0
65,536	0.27 GiB	1.63 GiB	5.99×	4 / 22 / 0
131,072	0.52 GiB	3.25 GiB	6.23×	4 / 22 / 0

22 of 26 layers cache only a 512-token window rather than the full context, on a period of 6. Figures assume the default configuration; --swa-full disables the saving entirely.

Compare with

same modality, comparable size

vs Qwen3-4B vs HyperCLOVAX-SEED-Text-Instruct-1.5B vs Llama-3.2-1B-Instruct vs ced-base vs gemma-3-1b-it vs Qwen3-1.7B 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 0.52 GiB. The real file is 0.75 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 0.81 GiB at 32K context where the real figure is 0.15 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

1152

Vocab

262,144

Sliding window

512

SWA period

MLA

Experts

—

Experts per token

—

use_sliding_window

—

Questions people ask

How much VRAM does next-1b need?: Q4_K_M is exactly 806,057,664 bytes (0.75 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
How large is next-1b's KV cache?: 0.15 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 next-1b 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.