Galactica TPS calculator

Open weights Meta AI 120B parameters November 2022

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

38 cards that can run it

818 cards we hold specifications for

Smallest card that fits

RTX PRO 5000 72 GB Blackwell

72 GB · Q3_K_M · 12.8 tok/s

Fastest card

H100 NVL 94 GB

32.1 tok/s · 94 GB

Which GPUs can run Galactica?

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

38 cards match

Calculating

					Needs	Quantisation	Fit
32.1 tok/s 19–51 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	73.3 GB	Q4_K_M	Tight
29.1 tok/s 17–47 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	66.3 GB	IQ4_XS	Tight
29.1 tok/s 17–47 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	66.3 GB	IQ4_XS	Tight
28.2 tok/s 17–45 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	129.2 GB	Q8_0	Comfortable
28.2 tok/s 17–45 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	129.2 GB	Q8_0	Comfortable
27.4 tok/s 16–44 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	73.3 GB	Q4_K_M	Tight
27.4 tok/s 16–44 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	73.3 GB	Q4_K_M	Tight
27.4 tok/s 16–44 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	73.3 GB	Q4_K_M	Tight
26.2 tok/s 16–42 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	101.2 GB	Q6_K	Tight
25.1 tok/s 15–40 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	101.2 GB	Q6_K	Comfortable
25.1 tok/s 15–40 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	101.2 GB	Q6_K	Comfortable
22.6 tok/s 14–36 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	129.2 GB	Q8_0	Comfortable
22.6 tok/s 14–36 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	129.2 GB	Q8_0	Comfortable
21.3 tok/s 13–34 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	101.2 GB	Q6_K	Tight
17.7 tok/s 11–28 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	66.3 GB	IQ4_XS	Tight
17.7 tok/s 11–28 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	66.3 GB	IQ4_XS	Tight
17.7 tok/s 11–28 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	66.3 GB	IQ4_XS	Tight
17.7 tok/s 11–28 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	66.3 GB	IQ4_XS	Tight
17.7 tok/s 11–28 · low confidence	H100 PCIe 80 GB NVIDIA	80 GB	2,040 GB/s	Oct 2022	66.3 GB	IQ4_XS	Tight
17.7 tok/s 11–28 · low confidence	H800 PCIe 80 GB NVIDIA	80 GB	2,040 GB/s	Mar 2023	66.3 GB	IQ4_XS	Tight
16.8 tok/s 10–27 · low confidence	A100 PCIe 80 GB NVIDIA	80 GB	1,940 GB/s	Jun 2021	66.3 GB	IQ4_XS	Tight
16.8 tok/s 10–27 · low confidence	A800 PCIe 80 GB NVIDIA	80 GB	1,940 GB/s	Nov 2022	66.3 GB	IQ4_XS	Tight
16.5 tok/s 10–26 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	129.2 GB	Q8_0	Comfortable
14.7 tok/s 9–23 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	129.2 GB	Q8_0	Comfortable
14.7 tok/s 9–23 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	129.2 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: Meta AI
Organisation type: Industry
Country: United States of America
Published: 16 November 2022
Authors: Ross Taylor, Marcin Kardas, Guillem Cucurull, Thomas Scialom, Anthony Hartshorn, Elvis Saravia, Andrew Poulton, Viktor Kerkez, Robert Stojnic

What it does

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

Domain: Language, Biology
Task: Language modeling, Question answering, Mathematical reasoning, Medical diagnosis, Language modeling/generation
Approach: Self-supervised learning

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: 120B
Training data: 106,000,000,000 tokens
Epochs: 4
Batch size: 2,000,000

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: 3.2 × 10²³ 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: NVIDIA A100 SXM4 80 GB
Chips used: 128
Power draw: 102.4 kW
Compute cost: $591,077

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 (non-commercial)
Training code: Unreleased

How it is classified

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

Foundation model: Yes
Likely above 10²³ FLOP: Yes
Why it is tracked: SOTA improvement
Record confidence: Likely
Citations: 1,017

Sources

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

Reference: Galactica: A Large Language Model for Science
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run Galactica

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 Galactica

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

Hardware requirements in practice

Minimum card

RTX PRO 5000 72 GB Blackwell

Memory needed

59.3 GB

Fastest

32.1 tok/s

Galactica sits at 120B parameters, which puts it above consumer hardware and into the range where a card is bought for this purpose rather than repurposed for it. 38 of the cards we track can hold it.

The entry point is the RTX PRO 5000 72 GB Blackwell: 72 GB of memory, Q3_K_M compression, roughly 12.8 tokens per second.

A H100 NVL 94 GB is the fastest we calculate for it: about 32.1 tokens per second, from 3,940 GB/s of memory bandwidth.

Where it came from

Galactica was published by Meta AI, in United States of America, in November 2022. The organisation is categorised as industry.

It works in Language, Biology, and is recorded as doing language modeling, Question answering, Mathematical reasoning, Medical diagnosis, Language modeling/generation.

The weights are published, so it can be downloaded and run on your own hardware indefinitely, offline, with no account attached.

Understanding the speeds

Half the cards that hold it manage more than 17.7 tokens per second, and 36 exceed reading speed outright.

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

Without the attention layout on record, the memory column is an approximation. It is close enough to choose hardware by, and least reliable at long context.

What went into building it

The training run consumed about 3.2 × 10²³ FLOP, on NVIDIA A100 SXM4 80 GB. That figure describes the cost of creating it and has no bearing on how quickly it generates text.

The training set ran to roughly 106,000,000,000 tokens.

The reason it appears in this catalogue at all is sOTA improvement.

Step by step

How to choose a GPU for Galactica

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

01

Check what it needs before anything else

Every card here has been checked against Galactica — around 59.3 GB at Q3_K_M. Capacity is the gate — a card either holds it or it does not.
02

Set the context length you will work at

Set the context to what you will actually use. The cache grows with the conversation, and it is the usual reason Galactica stops fitting a card that seemed fine.
03

Choose how far you will compress it

The quantisation column varies by card, because a bigger card holds a more accurate copy of Galactica — Q3_K_M on the smallest card that fits. Set a floor to hold the comparison at one level.
04

Sort by speed

Ranking by tokens per second for Galactica follows memory bandwidth, not core counts, which is why the H100 NVL 94 GB tops it at 32.1 tok/s.
05

Look at the headroom, not just the fit

The fit column separates cards that just manage Galactica from those with room to spare. Buy for the second if the context might grow.
06

Open the card you have settled on

Every card name links to its own page, which runs the same calculation across the whole model catalogue. Worth a look before buying for Galactica alone — a card is usually bought for more than one model.

Answers

Galactica — common questions

Can I run Galactica 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 Galactica is rarely worth using — the nearest miss we calculate is short by 15.7 GB. Every figure here assumes the whole model is on the card.

Would two GPUs run Galactica faster?

Two cards buy memory rather than speed. That matters for Galactica only if one card cannot hold it — 38 can, so a second adds little.

Why does the quantisation differ between cards for Galactica?

Because capacity varies, so does how hard Galactica has to be squeezed — 5 distinct levels appear in the table above. Set a minimum quality to compare at one.

How accurate are these Galactica speed estimates?

They are calculated from specifications rather than measured, and each carries a range — 19–51 tok/s on the H100 NVL 94 GB, for instance. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

What GPU do I need to run Galactica?

The smallest card in our catalogue that holds Galactica is the RTX PRO 5000 72 GB Blackwell, with 72 GB of memory. It runs the model at Q3_K_M using about 59.3 GB, and produces roughly 12.8 tokens per second. 38 cards in total can run it.

How fast is Galactica on a GPU?

It depends on the card. The quickest we calculate is a H100 NVL 94 GB at about 32.1 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 36 of the cards that can run Galactica clear that.

How much VRAM does Galactica need?

About 59.3 GB at Q3_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.

Is Galactica open source?

Its weights are published, so Galactica 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 Galactica have?

Galactica has 120B parameters. "The largest 120B model we train runs on a single NVIDIA A100 node". 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 Galactica?

Galactica was published by Meta AI, based in United States of America, categorised as industry.

When was Galactica released?

Galactica was published in November 2022. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

What is Galactica used for?

Galactica works in Language, Biology, and is recorded as handling language modeling, Question answering, Mathematical reasoning, Medical diagnosis, Language modeling/generation. These are the areas it was designed around; they describe intent rather than a hard boundary.

Where can I download Galactica?

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

How much compute was used to train Galactica?

Around 3.2 × 10²³ FLOP, on NVIDIA A100 SXM4 80 GB. 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 25 May 2026

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