Gemma 2B TPS calculator

Open weights Google DeepMind 2.5B parameters February 2024

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 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Tesla C1080

4 GB · Q8_0 · 14.7 tok/s

Fastest card

B200

1,352 tok/s · 180 GB

Which GPUs can run Gemma 2B?

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
1,352 tok/s 1,149–1,622	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	3.4 GB	Q8_0	Comfortable
1,352 tok/s 1,149–1,622	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	3.4 GB	Q8_0	Comfortable
1,079 tok/s 648–1,727 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	3.4 GB	Q8_0	Comfortable
1,079 tok/s 648–1,727 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	3.4 GB	Q8_0	Comfortable
863 tok/s 518–1,381 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	3.4 GB	Q8_0	Comfortable
826 tok/s 702–992	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	3.4 GB	Q8_0	Comfortable
826 tok/s 702–992	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	3.4 GB	Q8_0	Comfortable
791 tok/s 474–1,265 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	3.4 GB	Q8_0	Comfortable
702 tok/s 421–1,123 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	3.4 GB	Q8_0	Comfortable
702 tok/s 421–1,123 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	3.4 GB	Q8_0	Comfortable
702 tok/s 421–1,123 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	3.4 GB	Q8_0	Comfortable
666 tok/s 566–799	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	3.4 GB	Q8_0	Comfortable
568 tok/s 483–681	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	3.4 GB	Q8_0	Comfortable
568 tok/s 483–681	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	3.4 GB	Q8_0	Comfortable
568 tok/s 483–681	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	3.4 GB	Q8_0	Comfortable
568 tok/s 483–681	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	3.4 GB	Q8_0	Comfortable
568 tok/s 483–681	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	3.4 GB	Q8_0	Comfortable
432 tok/s 259–692 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	3.4 GB	Q8_0	Comfortable
432 tok/s 259–692 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	3.4 GB	Q8_0	Comfortable
360 tok/s 216–576 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	3.4 GB	Q8_0	Comfortable
353 tok/s 212–564 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	3.4 GB	Q8_0	Comfortable
345 tok/s 293–414	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	3.4 GB	Q8_0	Comfortable
345 tok/s 293–414	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	3.4 GB	Q8_0	Comfortable
345 tok/s 293–414	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	3.4 GB	Q8_0	Comfortable
345 tok/s 293–414	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	3.4 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: 21 February 2024
Authors: Gemma Team, Google DeepMind

What it does

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

Domain: Language
Task: Language modeling/generation, Chat, Code generation, Question answering, Quantitative reasoning

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: 2.5B
Training data: 3,000,000,000,000 tokens

Training compute

The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.

Training compute: 4.5 × 10²² FLOP
How it was established: Operation counting

The training run

What it physically took to train: which chips, how many, for how long, and what that drew from the wall.

Training hardware: Google TPU v5e
Chips used: 512
Power draw: 228.0 kW

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

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: Open Models Based on Gemini Research and Technology
Last updated: 28 November 2025

The extremes

The ten fastest GPUs that run Gemma 2B

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 2B

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

3.4 GB

Fastest

1,352 tok/s

Gemma 2B reaches a parameter count of 2.5B. That is small enough that hardware is rarely the obstacle, including on cards several years old. The number of cards we track that can run it: 818.

The smallest card that holds it is Tesla C1080, with a memory capacity of 4 GB, running it at a compression of Q8_0 and producing around 14.7 tokens per second.

At the other end sits B200, generating roughly 1,352 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

What this model is

Gemma 2B was published by Google DeepMind, in the country recorded as United States of America, during February 2024. It comes out of an organisation categorised as industry.

It works in the domain of Language, and is recorded as performing the task of language modeling/generation, Chat, Code generation, Question answering, Quantitative reasoning.

Published weights mean the model runs on your machine rather than someone else's, which is what makes the hardware question below answerable at all.

What decides the speed

Across every card that can run it, the middle of the range sits at 38.0 tokens per second. Clearing the ten tokens per second that roughly matches reading speed: 783 of them.

Being dense, it reads all of itself per token, which is why the ordering by speed below follows the ordering by memory bandwidth so closely.

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

What went into building it

The training run consumed about 4.5 × 10²² FLOP, on hardware recorded as Google TPU v5e. That figure measures what producing the model cost, and has no bearing on how fast it answers.

Training consumed a corpus of around 3,000,000,000,000 tokens of text.

Step by step

How to choose a GPU for Gemma 2B

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

Start from what it actually needs, which is the requirement of Gemma 2B, needing around 3.4 GB at a compression of Q8_0. No amount of processing power compensates for a card that cannot hold it.
02

Decide how long your conversations run

The conversation occupies memory too, and grows as it goes. Set the slider to the length you expect, because at long context a card that handles short questions easily can be dropped by Gemma 2B.
03

Decide how much compression you will accept

The quantisation column varies by card, because a bigger card holds a more accurate copy, reaching a compression of Q8_0 on the smallest card that fits. Setting a minimum quality drops the cards that only manage it by squeezing further than you would want, and holds the comparison at one level.
04

Compare tokens per second, not specifications

Sort by speed to see how cards rank for Gemma 2B. It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B200, at 1,352 tok/s.
05

Look at the headroom, not just the fit

Tight means it loads and works with no room to raise the context later, in the case of Gemma 2B. Comfortable means you can grow the context later. That difference matters more than a few tokens per second, so buy for comfortable if you expect to.
06

Check the card from the other side

Each card page repeats this sweep for every model we hold, answering what else the hardware is good for beyond Gemma 2B.

Answers

Gemma 2B — common questions

Gemma 2B— can I run it if it does not fit in my GPU?

Partly. Layers that do not fit sit in system memory and run at a fraction of the speed, so a mostly-offloaded model is rarely worth using. Every figure here assumes the whole model is resident on the card.

Gemma 2B— would two GPUs run it faster?

A second card roughly doubles the memory available but not the generation rate. The number already able to run it alone: 818. So a second card is rarely the answer here.

Gemma 2B— why does the quantisation differ between cards?

Each card is shown running the least-compressed copy it can hold. The number of distinct compression levels across the cards that fit it: 1. Bigger cards get the more accurate version, and the quality floor above pins the comparison to one level.

Gemma 2B— how accurate are these speed estimates?

They are calculated from specifications rather than measured, and each carries a range. One example: 1,149–1,622 tok/s on B200. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

Gemma 2B— what GPU do I need to run it?

The smallest card in our catalogue that holds it is Tesla C1080, with a memory capacity of 4 GB. It runs the model at a compression of Q8_0 using about 3.4 GB, and produces roughly 14.7 tokens per second. The number of cards able to run it in total: 818.

Gemma 2B— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B200, at about 1,352 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and the number of cards clearing that: 783.

Gemma 2B— how much VRAM does it need?

It needs about 3.4 GB at a compression of Q8_0, 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.

Gemma 2B— can I run it on a GPU holding 8 GB?

Yes. The card CMP 170HX 8 GB, holding 8 GB, runs it at a compression of Q8_0, using about 3.4 GB and generating roughly 252 tokens per second. The fit is comfortable.

Gemma 2B— can I run it on a GPU holding 12 GB?

Yes. The card GeForce RTX 3080 Ti, holding 12 GB, runs it at a compression of Q8_0, using about 3.4 GB and generating roughly 154 tokens per second. The fit is comfortable.

Gemma 2B— can I run it on a GPU holding 16 GB?

Yes. The card Tesla V100 SXM2 16 GB, holding 16 GB, runs it at a compression of Q8_0, using about 3.4 GB and generating roughly 191 tokens per second. The fit is comfortable.

Gemma 2B— can I run it on a GPU holding 24 GB?

Yes. The card GeForce RTX 5090 D V2, holding 24 GB, runs it at a compression of Q8_0, using about 3.4 GB and generating roughly 226 tokens per second. The fit is comfortable.

Gemma 2B— is it open source?

Its weights are published, so it 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.

Gemma 2B— how many parameters does it have?

It has a parameter count of 2.5B. Table 2, sum of embedding and non-embedding parameters: 2B 524,550,144 + 1,981,884,416 = 2506434560. 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.

Gemma 2B— who created it?

It was published by Google DeepMind, based in United States of America, an organisation categorised as industry.

Gemma 2B— when was it released?

It was published in February 2024. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

Gemma 2B— what is it used for?

It works in the domain of Language, and is recorded as handling the task of language modeling/generation, Chat, Code generation, Question answering, Quantitative reasoning. These are the areas it was designed around; they describe intent rather than a hard boundary.

Gemma 2B— where can I download it?

The weights are published, though we do not hold a repository link for it. This site calculates hardware requirements rather than hosting model files.

Gemma 2B— how much compute was used to train it?

Training consumed around 4.5 × 10²² FLOP, on hardware recorded as Google TPU v5e. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.

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