Gemma logo

Gemma 4 31B

Gemma 4 31B loads in 20 GB at Q4_K_M, comfortably from 36 GB of memory — Macs included, starting at the MacBook Pro M5 Pro 48GB. Below is the memory math we ran for it, the cheapest card that fits, and the fastest machine we track.

PARAMETERS
31B
FORMAT
Q4_K_M
MIN MEMORY
32 GB
BEST FOR
Quality, Coding, Multimodal

How much memory does Gemma 4 31B need?

Gemma 4 31B is a 31B dense model built for quality, coding, multimodal. The weights alone take 20 GB at Q4_K_M; add context and the OS, and it belongs in the 36 GB memory class. Every number below comes from the ModelFit engine — speeds are estimates, labeled est., because we do not benchmark hardware ourselves.

Weights (Q4_K_M)
20 GB
+ KV cache (16k)
~4.0 GB
Total at 16k
~24.0 GB
Comfortable from
36 GB

KV = fp16 estimate (q8_0 cache roughly halves it). "Comfortable" = weights + KV within the engine's tiered budget (~70-85% of memory).

Which machines run Gemma 4 31B?

Cheapest GPU
~28 tok/s est. — ~$700 used
Fastest
~46 tok/s est.
Runs on a Mac?
~13 tok/s est.

Verdicts come from the same engine that powers our fit checker — memory capacity first, bandwidth for speed. How we compute these numbers · Hugging Face model card

Go deeper

Every Gemma 4 31B quant by real GGUF file sizeAlso tracked: Gemma 4 31B (Q8) (Q8_0, 30.9 GB, min 48 GB)

More Gemma models

Frequently asked questions

How much memory does Gemma 4 31B need?

20 GB for the Q4_K_M weights, plus ~4.0 GB of KV-cache at 16k context — about 24.0 GB total. Comfortable from 36 GB of VRAM or unified memory.

Does Gemma 4 31B run on a Mac?

Yes — from the MacBook Pro M5 Pro 48GB (~13 tok/s est.). Unified memory means the RAM budget is the only limit.

What is the cheapest GPU for Gemma 4 31B?

The AMD Radeon RX 7900 XTX is the cheapest tracked card that runs Gemma 4 31B comfortably — ~28 tok/s est. at ~$700 used.

Which Gemma 4 31B quant should I pick?

The tracked Q4_K_M build at 20 GB is the one we recommend for most machines. Every published build, ranked by real GGUF file size rather than a formula, is on the quant comparison page linked below.

Cite this page

ModelFit: Gemma 4 31B — specs, memory math and hardware verdicts.
https://modelfit.io/models/gemma4-31b/ (dataset updated 2026-09-03, CC BY 4.0).