Gemma 3 QAT 4B TPS calculator

Open weights Google DeepMind 4B parameters April 2025

Each card below is assessed against this model at the context length and minimum quality you choose. Speed is an estimate for a single request, calculated from the card's memory bandwidth and the size of the model once compressed.

Calculated for this model

818 of 818 cards that can run it

Smallest card that fits

Tesla C1080

4 GB · Q4_K_M · 21.3 tok/s

Fastest card

B200

847 tok/s · 180 GB

Which GPUs can run Gemma 3 QAT 4B?

Set the inputs, read the answer

Context length tokens

A longer conversation needs more memory, which can push this model off smaller cards.

Minimum quality

Hides cards that would only fit the model by compressing it below this point.

Search cards

Card memory

Also show cards that cannot run it

818 cards match

Calculating

					Needs	Quantisation	Fit
847 tok/s 720–1,016	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	5.0 GB	Q8_0	Comfortable
847 tok/s 720–1,016	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	5.0 GB	Q8_0	Comfortable
676 tok/s 406–1,082 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	5.0 GB	Q8_0	Comfortable
676 tok/s 406–1,082 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	5.0 GB	Q8_0	Comfortable
541 tok/s 325–866 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	5.0 GB	Q8_0	Comfortable
518 tok/s 440–621	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	5.0 GB	Q8_0	Comfortable
518 tok/s 440–621	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	5.0 GB	Q8_0	Comfortable
496 tok/s 297–793 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	5.0 GB	Q8_0	Comfortable
440 tok/s 264–704 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	5.0 GB	Q8_0	Comfortable
440 tok/s 264–704 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	5.0 GB	Q8_0	Comfortable
440 tok/s 264–704 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	5.0 GB	Q8_0	Comfortable
417 tok/s 355–501	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	5.0 GB	Q8_0	Comfortable
356 tok/s 302–427	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	5.0 GB	Q8_0	Comfortable
356 tok/s 302–427	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	5.0 GB	Q8_0	Comfortable
356 tok/s 302–427	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	5.0 GB	Q8_0	Comfortable
356 tok/s 302–427	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	5.0 GB	Q8_0	Comfortable
356 tok/s 302–427	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	5.0 GB	Q8_0	Comfortable
271 tok/s 163–433 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	5.0 GB	Q8_0	Comfortable
271 tok/s 163–433 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	5.0 GB	Q8_0	Comfortable
226 tok/s 135–361 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	5.0 GB	Q8_0	Comfortable
221 tok/s 133–353 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	5.0 GB	Q8_0	Comfortable
216 tok/s 184–259	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	5.0 GB	Q8_0	Comfortable
216 tok/s 184–259	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	5.0 GB	Q8_0	Comfortable
216 tok/s 184–259	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	5.0 GB	Q8_0	Comfortable
216 tok/s 184–259	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	5.0 GB	Q8_0	Comfortable

Speeds are estimates for a single request — one conversation at a time — calculated from memory bandwidth, model size and quantisation. Real throughput varies with the inference runtime and its version. Figures published by hardware vendors measure many simultaneous requests and are much higher.

On record

Full specification

Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.

Origin

Who built this model, where, and when it was published.

Organisation: Google DeepMind
Organisation type: Industry
Country: United States of America
Published: 18 April 2025

What it does

The problem areas the model was built for. A model can carry several of each.

Domain: Language, Vision, Multimodal
Task: Language modeling/generation, Question answering, Translation, Chat, Quantitative reasoning, Visual question answering, Code generation
Base model: Gemma 3 4B

Size

How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.

Parameters: 4B
Training data: tokens

Availability

Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.

Weights: Open — downloadable
Model access: Open weights (restricted use)
Training code: Unreleased
Hugging Face: google

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Record confidence: Confident

Sources

Where this record came from and when it was last checked.

Reference: Gemma 3 QAT Models: Bringing state-of-the-Art AI to consumer GPUs
Last updated: 28 November 2025

The extremes

The ten fastest GPUs for Gemma 3 QAT 4B

Ranked by estimated tokens per second, newest card first where speeds tie. Because generation is bound by memory bandwidth, this ordering follows bandwidth rather than any gaming benchmark.

The smallest GPUs that still run Gemma 3 QAT 4B

The cheapest route in, by memory capacity. A tight fit runs the model but leaves nothing spare for a longer conversation.

What the numbers mean

What it takes to run this model

Minimum card

Tesla C1080

Memory needed

3.2 GB

Fastest

847 tok/s

Gemma 3 QAT 4B is small enough at 4B parameters that hardware is rarely the obstacle — 818 of the cards we track can run it, including cards several years old.

The smallest card that holds it is the Tesla C1080 with 4 GB, running it at Q4_K_M and producing around 21.3 tokens per second.

Top of the range is the B200, at roughly 847 tokens per second thanks to 8,000 GB/s of bandwidth.

Background

Gemma 3 QAT 4B was published by Google DeepMind, in United States of America, in April 2025. It comes out of industry.

It works in Language, Vision, Multimodal, and is recorded as doing language modeling/generation, Question answering, Translation, Chat, Quantitative reasoning, Visual question answering, Code generation.

It is derived from Gemma 3 4B rather than trained from scratch, which is the usual way a specialised model is produced.

The weights being open is what puts this page in the calculator rather than only in the catalogue: it is a model you can actually hold. It is published under the google organisation on Hugging Face.

Reading the throughput figures

The median result is around 31.3 tokens per second; 779 cards produce text faster than most people read it.

Every weight participates in every token here, so bandwidth is the whole story: the ranking below is effectively a ranking of memory throughput.

Its attention layout is on file, so the memory figures are computed exactly rather than approximated.

Step by step

How to choose a GPU for Gemma 3 QAT 4B

The table above has already assessed every card we hold specifications for against this model. Getting to your answer takes six steps.

01

Read the memory figure first

Look at what Gemma 3 QAT 4B actually needs — around 3.2 GB at Q4_K_M. No amount of processing power compensates for a card that cannot hold it.
02

Match the context to your actual use

The conversation occupies memory too, and grows as it goes. Set the slider to the length you expect: at long context Gemma 3 QAT 4B can slip off a card that handles short questions easily.
03

Decide how much compression you will accept

The quantisation column varies by card, because a bigger card holds a more accurate copy of Gemma 3 QAT 4B — Q4_K_M on the smallest card that fits. Set a floor to hold the comparison at one level.
04

Compare tokens per second, not specifications

The speed ordering for Gemma 3 QAT 4B is effectively an ordering by memory bandwidth, which is why the B200 tops it at 847 tok/s.
05

Look at the headroom, not just the fit

A tight fit runs Gemma 3 QAT 4B but leaves nothing spare for a longer conversation; comfortable has headroom. If you expect to grow the context, buy for comfortable.
06

Check the card from the other side

Each card page repeats this sweep for every model we hold. It answers what else the hardware is good for, beyond Gemma 3 QAT 4B.

Answers

Gemma 3 QAT 4B — common questions

Where can I download Gemma 3 QAT 4B?

Its weights are published under the google organisation on Hugging Face. We do not host model files — this site calculates what hardware is needed to run them.

Can I run Gemma 3 QAT 4B if it does not fit in my GPU?

It can be split between the card and system memory, but Gemma 3 QAT 4B generates painfully slowly that way. Nothing on this page assumes offloading.

Would two GPUs run Gemma 3 QAT 4B faster?

Two cards buy memory rather than speed. That matters for Gemma 3 QAT 4B only if one card cannot hold it — 818 can, so a second adds little.

Why does the quantisation differ between cards for Gemma 3 QAT 4B?

Because capacity varies, so does how hard Gemma 3 QAT 4B has to be squeezed — 3 distinct levels appear in the table above. Set a minimum quality to compare at one.

How accurate are these Gemma 3 QAT 4B speed estimates?

These are estimates with real error bars. The fastest result here, 720–1,016 tok/s on the B200, could reasonably land anywhere in its published range depending on which runtime you use.

What GPU do I need to run Gemma 3 QAT 4B?

The smallest card in our catalogue that holds Gemma 3 QAT 4B is the Tesla C1080, with 4 GB of memory. It runs the model at Q4_K_M using about 3.2 GB, and produces roughly 21.3 tokens per second. 818 cards in total can run it.

How fast is Gemma 3 QAT 4B on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 847 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 779 of the cards that can run Gemma 3 QAT 4B clear that.

How much VRAM does Gemma 3 QAT 4B need?

About 3.2 GB at Q4_K_M compression, which is what the smallest card that runs it uses. Less compression needs more: the figures in the memory column above are recalculated for each card, because each one holds the least-compressed version it can.

Can I run Gemma 3 QAT 4B on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q8_0, using about 5.0 GB and generating roughly 158 tokens per second — a comfortable fit.

Can I run Gemma 3 QAT 4B on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q8_0, using about 5.0 GB and generating roughly 96.6 tokens per second — a comfortable fit.

Can I run Gemma 3 QAT 4B on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 5.0 GB and generating roughly 120 tokens per second — a comfortable fit.

Can I run Gemma 3 QAT 4B on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 5.0 GB and generating roughly 142 tokens per second — a comfortable fit.

Is Gemma 3 QAT 4B open source?

Its weights are published, so Gemma 3 QAT 4B can be downloaded and run on your own hardware. Note that open weights is not the same as open source in the full sense — it says nothing about the training data, the training code, or the commercial terms attached.

How many parameters does Gemma 3 QAT 4B have?

Gemma 3 QAT 4B has 4B parameters. 4B. That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.

Who created Gemma 3 QAT 4B?

Gemma 3 QAT 4B was published by Google DeepMind, based in United States of America, categorised as industry.

When was Gemma 3 QAT 4B released?

Gemma 3 QAT 4B was published in April 2025.

What is Gemma 3 QAT 4B used for?

Gemma 3 QAT 4B works in Language, Vision, Multimodal, and is recorded as handling language modeling/generation, Question answering, Translation, Chat, Quantitative reasoning, Visual question answering, Code generation. These are the areas it was designed around; they describe intent rather than a hard boundary.

Source

Original publication

Record last updated 28 November 2025

The other direction

Looking at it from the other side?

This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.

All AI models