Gemma 4 31B hardware requirements

What is Gemma 4 31B?

Gemma 4 31B is the largest model in Google DeepMind's Gemma 4 family, released in April 2026. It is a dense model that reads text and images with up to a 256k context, built for reasoning, coding and agent work with native function calling. Its thinking mode is off by default, and unlike Gemma 3 it is licensed under Apache 2.0.

Model card · Release post

What to know before running Gemma 4 31B

Gemma 4 31B is a dense model, so every token reads all of its weights. At Q4_K_M it needs 19.5 GB with an 8k context, which makes 24GB the size that holds it completely: the GeForce RTX 3090, the GeForce RTX 4090 and the Radeon RX 7900 XTX all run it great. On 16GB cards part of it has to run from system RAM, and it is slow; the IQ4_XS file does not fit in 16GB either.

If you have less than 24GB, its sibling Gemma 4 26B-A4B is a mixture-of-experts model that uses about 4B parameters per token and handles small cards far better; on the GeForce RTX 5060 Ti 16GB CanRun expects it to run well, a theoretical estimate (±30%). Gemma 4 31B reads images as well as text. Its thinking mode is off by default, but CanRun still holds models that have one to a 1.5× stricter speed bar, so with thinking left off it may do a grade better than the table shows. On a Mac, the Apple M4 Pro needs its 48GB configuration to keep it on the GPU and runs it slowly, while the faster Apple M4 Max (40-core GPU) runs it well.

About the model

Parameters: 31.27B
Attention: Sliding-window attention
Max context: 262,144 tokens
Released: 2026-03-11
Model card: Gemma 4 31B on Hugging Face

Other sizes in this series

Graphics card memory each model needs at its baseline quant with 8k context, and its verdict on common graphics cards, smallest model first
Model	Parameters	Baseline quant	Memory needed (8k)	GeForce RTX 4060	GeForce RTX 5060 Ti 16GB	GeForce RTX 3090
Gemma 4 E2B	5.12B	Q4_K_M	2.1 GB	Runs great	Runs great	Runs great
Gemma 4 E4B	8B	Q4_K_M	3.7 GB	Runs great	Runs great	Runs great
Gemma 4 12B	11.96B	Q4_K_M	7.8 GB	Runs well	Runs great	Runs great
Gemma 4 26B-A4B	26.54B total · 4B active (MoE)	Q4_K_M	17.7 GB	Runs well	Runs well	Runs great
Gemma 4 31Bthis model	31.27B	Q4_K_M	19.5 GB	Runs slowly	Runs slowly	Runs great

GGUF files

Tracked quantizations of Gemma 4 31B, smallest first
Quant	File size	Quality	Published by
UD-Q2_K_XL	11.77 GB	Heavy loss	Community GGUF · unsloth
IQ4_XS	16.37 GB	Good balance	Community GGUF · unsloth
Q4_K_Mbaseline	18.32 GB	Good balance	Community GGUF · unsloth
Q8_0	32.64 GB	Near-lossless	Community GGUF · unsloth

File sizes do not include the vision projector (mmproj) that image input needs.

Some repositories publish only Unsloth Dynamic (UD) variants; sizes are those of the file you would download.

Memory needed by quant and context

Graphics card memory in GB: weights loaded on the GPU + f16 KV cache + compute buffer. Add about 0.6 GB if the GPU also drives your display on Windows.
Quant	4k	8k	16k	32k	64k	128k	256k
UD-Q2_K_XL	12.6	13.0	13.7	15.2	18.2	24.2	36.1
IQ4_XS	17.2	17.6	18.3	19.8	22.8	28.8	40.7
Q4_K_M	19.2	19.5	20.3	21.8	24.8	30.7	42.7
Q8_0	33.5	33.9	34.6	36.1	39.1	45.1	57.0

Which GPUs can run Gemma 4 31B?

How to read the verdicts

Runs great: Fully on the GPU at 20 tok/s or more
Runs well: Fully on the GPU at 8–20 tok/s; MoE experts in system RAM at 20 tok/s or more; or at least 90% on the GPU at 8 tok/s or more
Runs slowly: 2–8 tok/s; CPU-only; less than 90% on the GPU; or MoE experts in system RAM below 20 tok/s
Won't run: Does not fit, or under 2 tok/s

Dense models split between GPU and CPU slow down sharply — the CPU side sets the pace. MoE models that keep only their experts in system RAM degrade far more gently.

Sort

Filter this list by budget, type and context in the calculator

Scroll sideways to see every column.

58 GPUs, Macs and CPU setups at Q4_K_M with 8k context, sorted by verdict and speed
Hardware	Verdict	Speed	Quant	Memory	Runs as	Context	Notes	Price
GeForce RTX 5090	Runs greatGemma 4 31B verdict on the GeForce RTX 5090	est. 66.1 tok/s58.1–74.0calibrated estimate ±12%	Q4_K_M	20.1 / 32.0 GB	Full GPU	up to 128k	—	$4,200 street (as of 2026-08-09)
NVIDIA RTX PRO 6000 Blackwell	Runs great	est. 66.1 tok/s58.1–74.0calibrated estimate ±12%	Q4_K_M	20.1 / 96.0 GB	Full GPU	up to 256k	—	—
NVIDIA RTX PRO 6000 Blackwell Max-Q	Runs great	est. 66.1 tok/s58.1–74.0calibrated estimate ±12%	Q4_K_M	20.1 / 96.0 GB	Full GPU	up to 256k	—	—
Radeon RX 7900 XTX	Runs great	est. 38.4 tok/s33.8–43.0calibrated estimate ±12%	Q4_K_M	20.1 / 24.0 GB	Full GPU	up to 32k	—	$999 launch MSRP
GeForce RTX 3090 Ti	Runs great	est. 37.2 tok/s32.7–41.6calibrated estimate ±12%	Q4_K_M	20.1 / 24.0 GB	Full GPU	up to 32k	—	$1,999 launch MSRP
GeForce RTX 4090	Runs greatGemma 4 31B verdict on the GeForce RTX 4090	est. 37.2 tok/s32.7–41.6calibrated estimate ±12%	Q4_K_M	20.1 / 24.0 GB	Full GPU	up to 32k	—	$2,755 street (as of 2026-08-09)
Apple M5 Ultra · 96 GB	Runs great	est. 36.9 tok/s29.5–44.2calibrated estimate ±20%	Q4_K_M	19.6 / 67.2 GB	Unified memory	up to 32k	—	—
GeForce RTX 3090	Runs greatGemma 4 31B verdict on the GeForce RTX 3090	est. 34.5 tok/s30.4–38.6calibrated estimate ±12%	Q4_K_M	20.1 / 24.0 GB	Full GPU	up to 32k	—	$1,050 street (as of 2026-08-09)
GeForce RTX 5090 Laptop	Runs great	est. 33.0 tok/s29.1–37.0calibrated estimate ±12%	Q4_K_M	20.1 / 24.0 GB	Full GPU	up to 16k	—	—
Radeon RX 7900 XT	Runs well	est. 28.7 tok/s23.0–34.5calibrated estimate ±20%	Q4_K_M	20.0 / 20.0 GB + 0.2 GB RAM	Partial offload	up to 16k	Try IQ4_XS: Runs great	$899 launch MSRP
Apple M3 Ultra · 96 GB	Runs well	est. 24.6 tok/s19.7–29.5calibrated estimate ±20%	Q4_K_M	19.6 / 67.2 GB	Unified memory	up to 128k	UD-Q2_K_XL (heavy quality loss): Runs great	—
Apple M5 Max (40-core GPU) · 48 GB	Runs well	est. 24.6 tok/s19.7–29.5calibrated estimate ±20%	Q4_K_M	19.6 / 33.6 GB	Unified memory	up to 128k	UD-Q2_K_XL (heavy quality loss): Runs great	—
Apple M2 Ultra · 64 GB	Runs well	est. 24.0 tok/s19.2–28.8calibrated estimate ±20%	Q4_K_M	19.6 / 44.8 GB	Unified memory	up to 128k	UD-Q2_K_XL (heavy quality loss): Runs great	—
Apple M4 Max (40-core GPU) · 48 GB	Runs well	est. 21.9 tok/s17.5–26.2calibrated estimate ±20%	Q4_K_M	19.6 / 33.6 GB	Unified memory	up to 128k	UD-Q2_K_XL (heavy quality loss): Runs great	—
Apple M5 Max (32-core GPU) · 36 GB	Runs well	est. 18.4 tok/s14.7–22.1calibrated estimate ±20%	Q4_K_M	19.6 / 25.2 GB	Unified memory	up to 64k	—	—
Apple M4 Max (32-core GPU) · 36 GB	Runs well	est. 16.4 tok/s13.1–19.7calibrated estimate ±20%	Q4_K_M	19.6 / 25.2 GB	Unified memory	up to 64k	—	—
Apple M5 Pro · 48 GB	Runs well	est. 12.3 tok/s9.8–14.7calibrated estimate ±20%	Q4_K_M	19.6 / 33.6 GB	Unified memory	up to 8k	—	—
Ryzen AI Max+ 395 (Strix Halo) · 32 GB	Runs slowly	est. 11.1 tok/s8.8–13.3calibrated estimate ±20%	Q4_K_M	19.6 / 22.4 GB	Unified memory	up to 32k	Try IQ4_XS: Runs well	$1,999 launch MSRP (128 GB)
NVIDIA DGX Spark · 128 GB	Runs slowly	10.1 tok/s8k estimate 11.2 tok/s (9.0–13.5, calibrated estimate ±20%)measured (1 run, 512 context)	Q4_K_M	19.6 / 116.9 GB	Unified memory	up to 256k	Try IQ4_XS: Runs well	$3,999 launch MSRP (128 GB)
GeForce RTX 5080	Runs slowly	est. 8.5 tok/s6.8–10.1calibrated estimate ±20%	Q4_K_M	16.0 / 16.0 GB + 4.2 GB RAM	Partial offload	up to 64k	UD-Q2_K_XL (heavy quality loss): Runs great	$1,256 street (as of 2026-08-09)
GeForce RTX 5070 Ti	Runs slowly	est. 8.3 tok/s6.7–10.0calibrated estimate ±20%	Q4_K_M	16.0 / 16.0 GB + 4.2 GB RAM	Partial offload	up to 64k	UD-Q2_K_XL (heavy quality loss): Runs great	$949 street (as of 2026-08-09)
GeForce RTX 5080 Laptop	Runs slowly	est. 8.3 tok/s6.7–10.0calibrated estimate ±20%	Q4_K_M	16.0 / 16.0 GB + 4.2 GB RAM	Partial offload	up to 64k	UD-Q2_K_XL (heavy quality loss): Runs great	—
GeForce RTX 4080 Super	Runs slowly	est. 8.0 tok/s6.4–9.6calibrated estimate ±20%	Q4_K_M	16.0 / 16.0 GB + 4.2 GB RAM	Partial offload	up to 64k	UD-Q2_K_XL (heavy quality loss): Runs great	$999 launch MSRP
GeForce RTX 4080	Runs slowly	est. 8.0 tok/s6.4–9.6calibrated estimate ±20%	Q4_K_M	16.0 / 16.0 GB + 4.2 GB RAM	Partial offload	up to 64k	UD-Q2_K_XL (heavy quality loss): Runs great	$1,199 launch MSRP
Radeon RX 9070	Runs slowly	est. 7.9 tok/s6.3–9.5calibrated estimate ±20%	Q4_K_M	16.0 / 16.0 GB + 4.2 GB RAM	Partial offload	up to 64k	UD-Q2_K_XL (heavy quality loss): Runs great	$639 street (as of 2026-08-09)
Radeon RX 9070 XT	Runs slowly	est. 7.9 tok/s6.3–9.5calibrated estimate ±20%	Q4_K_M	16.0 / 16.0 GB + 4.2 GB RAM	Partial offload	up to 64k	UD-Q2_K_XL (heavy quality loss): Runs great	$689 street (as of 2026-08-09)
Radeon RX 7800 XT	Runs slowly	est. 7.9 tok/s6.3–9.4calibrated estimate ±20%	Q4_K_M	16.0 / 16.0 GB + 4.2 GB RAM	Partial offload	up to 64k	UD-Q2_K_XL (heavy quality loss): Runs great	$499 launch MSRP
GeForce RTX 4070 Ti Super	Runs slowly	est. 7.8 tok/s6.3–9.4calibrated estimate ±20%	Q4_K_M	16.0 / 16.0 GB + 4.2 GB RAM	Partial offload	up to 64k	UD-Q2_K_XL (heavy quality loss): Runs great	$799 launch MSRP
GeForce RTX 4090 Laptop	Runs slowly	est. 7.5 tok/s6.0–9.0calibrated estimate ±20%	Q4_K_M	16.0 / 16.0 GB + 4.2 GB RAM	Partial offload	up to 64k	UD-Q2_K_XL (heavy quality loss): Runs great	—
Apple M4 Pro · 24 GB	Runs slowlyGemma 4 31B verdict on the Apple M4 Pro	est. 7.2 tok/s5.0–9.3theoretical estimate ±30%	Q4_K_M	19.0 GB RAM	CPU only	up to 16k	UD-Q2_K_XL (heavy quality loss): Runs well	—
GeForce RTX 5060 Ti 16GB	Runs slowlyGemma 4 31B verdict on the GeForce RTX 5060 Ti 16GB	est. 7.0 tok/s5.6–8.4calibrated estimate ±20%	Q4_K_M	16.0 / 16.0 GB + 4.2 GB RAM	Partial offload	up to 64k	UD-Q2_K_XL (heavy quality loss): Runs well	$569 street (as of 2026-08-09)
Radeon RX 9060 XT 16GB	Runs slowly	est. 6.4 tok/s5.1–7.7calibrated estimate ±20%	Q4_K_M	16.0 / 16.0 GB + 4.2 GB RAM	Partial offload	up to 64k	UD-Q2_K_XL (heavy quality loss): Runs well	$449 street (as of 2026-08-09)
GeForce RTX 4060 Ti 16GB	Runs slowly	est. 5.9 tok/s4.7–7.1calibrated estimate ±20%	Q4_K_M	16.0 / 16.0 GB + 4.2 GB RAM	Partial offload	up to 64k	UD-Q2_K_XL (heavy quality loss): Runs well	$499 launch MSRP
GeForce RTX 3080 12GB	Runs slowly	est. 4.9 tok/s3.9–5.8calibrated estimate ±20%	Q4_K_M	12.0 / 12.0 GB + 8.2 GB RAM	Partial offload	up to 32k	—	$799 launch MSRP
GeForce RTX 5070	Runs slowlyGemma 4 31B verdict on the GeForce RTX 5070	est. 4.7 tok/s3.8–5.7calibrated estimate ±20%	Q4_K_M	12.0 / 12.0 GB + 8.2 GB RAM	Partial offload	up to 32k	—	$629 street (as of 2026-08-09)
GeForce RTX 5070 Ti Laptop	Runs slowly	est. 4.7 tok/s3.8–5.7calibrated estimate ±20%	Q4_K_M	12.0 / 12.0 GB + 8.2 GB RAM	Partial offload	up to 32k	—	—
GeForce RTX 4070	Runs slowlyGemma 4 31B verdict on the GeForce RTX 4070	est. 4.6 tok/s3.7–5.5calibrated estimate ±20%	Q4_K_M	12.0 / 12.0 GB + 8.2 GB RAM	Partial offload	up to 32k	—	$599 launch MSRP
GeForce RTX 4070 Super	Runs slowly	est. 4.6 tok/s3.7–5.5calibrated estimate ±20%	Q4_K_M	12.0 / 12.0 GB + 8.2 GB RAM	Partial offload	up to 32k	—	$599 launch MSRP
Apple M6 · 24 GB	Runs slowly	est. 4.5 tok/s3.1–5.8theoretical estimate ±30%	Q4_K_M	19.0 GB RAM	CPU only	up to 16k	—	—
GeForce RTX 4080 Laptop	Runs slowly	est. 4.5 tok/s3.6–5.4calibrated estimate ±20%	Q4_K_M	12.0 / 12.0 GB + 8.2 GB RAM	Partial offload	up to 32k	—	—
Intel Arc B580	Runs slowly	est. 4.4 tok/s3.5–5.3calibrated estimate ±20%	Q4_K_M	12.0 / 12.0 GB + 8.2 GB RAM	Partial offload	up to 32k	—	$290 street (as of 2026-08-09)
GeForce RTX 3060 12GB	Runs slowlyGemma 4 31B verdict on the GeForce RTX 3060 12GB	est. 4.3 tok/s3.5–5.2calibrated estimate ±20%	Q4_K_M	12.0 / 12.0 GB + 8.2 GB RAM	Partial offload	up to 32k	—	$250 street (as of 2026-08-09)
GeForce RTX 2080 Ti	Runs slowly	est. 4.3 tok/s3.4–5.1calibrated estimate ±20%	Q4_K_M	11.0 / 11.0 GB + 9.2 GB RAM	Partial offload	up to 32k	—	$999 launch MSRP
GeForce GTX 1080 Ti	Runs slowly	est. 4.2 tok/s2.9–5.4theoretical estimate ±30%	Q4_K_M	11.0 / 11.0 GB + 9.2 GB RAM	Partial offload	up to 32k	—	$699 launch MSRP
Apple M5 · 24 GB	Runs slowly	est. 4.0 tok/s2.8–5.3theoretical estimate ±30%	Q4_K_M	19.0 GB RAM	CPU only	up to 16k	—	—
GeForce RTX 3080 10GB	Runs slowly	est. 4.0 tok/s3.2–4.8calibrated estimate ±20%	Q4_K_M	10.0 / 10.0 GB + 10.2 GB RAM	Partial offload	up to 32k	—	$699 launch MSRP
GeForce RTX 3070	Runs slowly	est. 3.3 tok/s2.7–4.0calibrated estimate ±20%	Q4_K_M	8.0 / 8.0 GB + 12.2 GB RAM	Partial offload	up to 16k	—	$499 launch MSRP
GeForce RTX 5060	Runs slowly	est. 3.3 tok/s2.7–4.0calibrated estimate ±20%	Q4_K_M	8.0 / 8.0 GB + 12.2 GB RAM	Partial offload	up to 16k	—	$339 street (as of 2026-08-09)
GeForce RTX 5060 Ti 8GB	Runs slowly	est. 3.3 tok/s2.7–4.0calibrated estimate ±20%	Q4_K_M	8.0 / 8.0 GB + 12.2 GB RAM	Partial offload	up to 16k	—	$429 street (as of 2026-08-09)
GeForce RTX 4060 Ti 8GB	Runs slowly	est. 3.2 tok/s2.6–3.8calibrated estimate ±20%	Q4_K_M	8.0 / 8.0 GB + 12.2 GB RAM	Partial offload	up to 8k	—	$399 launch MSRP
GeForce RTX 4060	Runs slowlyGemma 4 31B verdict on the GeForce RTX 4060	est. 3.2 tok/s2.5–3.8calibrated estimate ±20%	Q4_K_M	8.0 / 8.0 GB + 12.2 GB RAM	Partial offload	up to 8k	—	$299 launch MSRP
Apple M4 · 24 GB	Runs slowlyGemma 4 31B verdict on the Apple M4	est. 3.2 tok/s2.2–4.1theoretical estimate ±30%	Q4_K_M	19.0 GB RAM	CPU only	up to 16k	—	—
GeForce RTX 4060 Laptop	Runs slowly	est. 3.2 tok/s2.5–3.8calibrated estimate ±20%	Q4_K_M	8.0 / 8.0 GB + 12.2 GB RAM	Partial offload	up to 8k	—	—
GeForce RTX 4070 Laptop	Runs slowly	est. 3.2 tok/s2.5–3.8calibrated estimate ±20%	Q4_K_M	8.0 / 8.0 GB + 12.2 GB RAM	Partial offload	up to 8k	—	—
GeForce RTX 3050 8GB	Runs slowly	est. 3.1 tok/s2.5–3.7calibrated estimate ±20%	Q4_K_M	8.0 / 8.0 GB + 12.2 GB RAM	Partial offload	up to 8k	—	$249 launch MSRP
DDR5-6000 dual-channel CPU · 96 GB	Won't runTry UD-Q2_K_XL (heavy quality loss): Runs slowly	est. 3.9 tok/s3.1–4.6calibrated estimate ±20%	UD-Q2_K_XL	12.4 GB RAM	CPU only	up to 32k	—	—
DDR5-5600 dual-channel CPU · 96 GB	Won't runTry UD-Q2_K_XL (heavy quality loss): Runs slowly	est. 3.6 tok/s2.9–4.3calibrated estimate ±20%	UD-Q2_K_XL	12.4 GB RAM	CPU only	up to 32k	—	—
DDR4-3200 dual-channel CPU · 64 GB	Won't run	est. 1.3 tok/s1.1–1.6calibrated estimate ±20%	Q4_K_M	19.0 GB RAM	CPU only	—	Fits, but too slow to use	—

Models with a thinking mode spend extra tokens thinking, so their speed thresholds are 1.5× stricter (30 / 12 / 3 tok/s). With thinking turned off, a model may do one grade better.

Cheapest GPUs that run it

Runs greatGeForce RTX 3090$1,050 street (as of 2026-08-09), est. 34.5 tok/s (30.4–38.6, calibrated estimate ±12%).

Measured results for Gemma 4 31B

Public benchmarks we calibrate against. Their conditions (context, backend, flags) can differ from the estimates above.

Measured results for Gemma 4 31B
Hardware	Quant	Backend	Context	Prompt (tok/s)	Generation (tok/s)	Flags	Source	Measured
Ryzen AI Max+ 395 (Strix Halo)	UD-Q4_K_XL	llama.cpp	512	302.0	11.5	Vulkan	github.com	2026-05-18
NVIDIA DGX Spark	Q4_K_M	llama.cpp	512	687.0	10.1	CUDA	github.com	2026-04-14

Frequently asked questions

How much VRAM does Gemma 4 31B need?

At Q4_K_M the file is 18.32 GB; with 8k context a graphics card needs about 19.5 GB (weights 18.30 GB, KV cache 0.7 GB, compute buffer 0.6 GB). The smallest tracked file (UD-Q2_K_XL (heavy quality loss)) needs about 13.0 GB.

What is the cheapest GPU that runs Gemma 4 31B well?

Cheapest that runs it great: GeForce RTX 3090 at $1,050 (street price as of 2026-08-09), est. 34.5 tok/s (30.4–38.6, calibrated estimate ±12%).

Can Gemma 4 31B run on an 8 GB, 16 GB or 24 GB GPU?

At Q4_K_M with 8k context: GeForce RTX 4060 — Runs slowly, est. 3.2 tok/s (2.5–3.8, calibrated estimate ±20%); GeForce RTX 5060 Ti 16GB — Runs slowly, est. 7.0 tok/s (5.6–8.4, calibrated estimate ±20%); GeForce RTX 3090 — Runs great, est. 34.5 tok/s (30.4–38.6, calibrated estimate ±12%).

Speeds are estimates from memory bandwidth, calibrated against public benchmarks, and each one comes with an error band and a confidence label. Real results vary with drivers, backend, context length and thermals.