BERT-Large TPS calculator

Open weights Google 340M parameters October 2018

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 · 108 tok/s

Fastest card

B200

9,965 tok/s · 180 GB

Which GPUs can run BERT-Large?

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
9,965 tok/s 5,979–15,945 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	1.1 GB	Q8_0	Comfortable
9,965 tok/s 5,979–15,945 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	1.1 GB	Q8_0	Comfortable
7,958 tok/s 4,775–12,732 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	1.1 GB	Q8_0	Comfortable
7,958 tok/s 4,775–12,732 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	1.1 GB	Q8_0	Comfortable
6,364 tok/s 3,818–10,183 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	1.1 GB	Q8_0	Comfortable
6,091 tok/s 3,655–9,746 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	1.1 GB	Q8_0	Comfortable
6,091 tok/s 3,655–9,746 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	1.1 GB	Q8_0	Comfortable
5,830 tok/s 3,498–9,328 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	1.1 GB	Q8_0	Comfortable
5,174 tok/s 3,104–8,278 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	1.1 GB	Q8_0	Comfortable
5,174 tok/s 3,104–8,278 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	1.1 GB	Q8_0	Comfortable
5,174 tok/s 3,104–8,278 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	1.1 GB	Q8_0	Comfortable
4,908 tok/s 2,945–7,853 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	1.1 GB	Q8_0	Comfortable
4,185 tok/s 2,511–6,697 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	1.1 GB	Q8_0	Comfortable
4,185 tok/s 2,511–6,697 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	1.1 GB	Q8_0	Comfortable
4,185 tok/s 2,511–6,697 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	1.1 GB	Q8_0	Comfortable
4,185 tok/s 2,511–6,697 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	1.1 GB	Q8_0	Comfortable
4,185 tok/s 2,511–6,697 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	1.1 GB	Q8_0	Comfortable
3,187 tok/s 1,912–5,099 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	1.1 GB	Q8_0	Comfortable
3,187 tok/s 1,912–5,099 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	1.1 GB	Q8_0	Comfortable
2,656 tok/s 1,593–4,249 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	1.1 GB	Q8_0	Comfortable
2,599 tok/s 1,559–4,159 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	1.1 GB	Q8_0	Comfortable
2,541 tok/s 1,525–4,066 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	1.1 GB	Q8_0	Comfortable
2,541 tok/s 1,525–4,066 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	1.1 GB	Q8_0	Comfortable
2,541 tok/s 1,525–4,066 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	1.1 GB	Q8_0	Comfortable
2,541 tok/s 1,525–4,066 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	1.1 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
Organisation type: Industry
Country: United States of America
Published: 11 October 2018
Authors: J Devlin, MW Chang, K Lee, K Toutanova

What it does

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

Domain: Language
Task: Question answering, Text autocompletion
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: 340M
Training data: 2,649,900,000 tokens
Epochs: 40
Batch size: 128,000

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: 2.9 × 10²⁰ FLOP
How it was established: Operation counting,Hardware,Third-party estimation

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 v2
Chips used: 64
Chip-hours: 6,144
Wall-clock time: 96 hours
Hardware utilisation: MFU 28.0%
Power draw: 37.0 kW
Compute cost: $1,751

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 (unrestricted)
Training code: Open source

How it is classified

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

Why it is tracked: Highly cited
Record confidence: Confident
Citations: 114,811

Sources

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

Reference: BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run BERT-Large

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 BERT-Large

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

1.1 GB

Fastest

9,965 tok/s

BERT-Large reaches a parameter count of 340M. 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 108 tokens per second.

The fastest we calculate for it is B200, generating roughly 9,965 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

What this model is

BERT-Large was published by Google, in the country recorded as United States of America, during October 2018. It comes out of an organisation categorised as industry.

It works in the domain of Language, and is recorded as performing the task of question answering, Text autocompletion.

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.

What decides the speed

Across every card that can run it, the middle of the range sits at 279.8 tokens per second. Producing text faster than most people read it: 818 of them.

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.

Training and provenance

Producing it required arithmetic totalling around 2.9 × 10²⁰ FLOP, on hardware recorded as Google TPU v2. That figure measures what producing the model cost, and has no bearing on how fast it answers.

The training set ran to roughly 2,649,900,000 tokens of text.

Its inclusion criterion: highly cited.

Step by step

How to choose a GPU for BERT-Large

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

01

Start from the memory column

Start from what it actually needs, which is the requirement of BERT-Large, needing around 1.1 GB at a compression of Q8_0. That figure, not the headline performance of a card, is what decides whether it runs.
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, because at long context a card that handles short questions easily can be dropped by BERT-Large.
03

Set a quality floor

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

Sort by speed

The speed ordering is effectively an ordering by memory bandwidth, for BERT-Large. It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B200, at 9,965 tok/s.
05

Check the fit verdict before buying

Tight means it loads and works with no room to raise the context later, in the case of BERT-Large. 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

Every card name links to its own page, which runs the same calculation across the whole model catalogue. A card is usually bought for more than one model, so it is worth a look before buying for BERT-Large.

Answers

BERT-Large — common questions

BERT-Large— how many parameters does it have?

It has a parameter count of 340M. 340M. 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.

BERT-Large— who created it?

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

BERT-Large— when was it released?

It was published in October 2018. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

BERT-Large— what is it used for?

It works in the domain of Language, and is recorded as handling the task of question answering, Text autocompletion. These are the areas it was designed around; they describe intent rather than a hard boundary.

BERT-Large— 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.

BERT-Large— how much compute was used to train it?

Training consumed around 2.9 × 10²⁰ FLOP, on hardware recorded as Google TPU v2. 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.

BERT-Large— can I run it if it does not fit in my GPU?

Only by offloading, which is usually a false economy: the part held in system memory drags the whole thing down. Every figure here assumes the whole model is resident on the card.

BERT-Large— would two GPUs run it faster?

Capacity adds across cards; throughput does not. The number of cards already holding it on their own: 818. So a second card is rarely the answer here.

BERT-Large— why does the quantisation differ between cards?

Because capacity varies, so does how hard it has to be squeezed. The number of distinct levels in the table above: 1. Bigger cards get the more accurate version, and the quality floor above pins the comparison to one level.

BERT-Large— how accurate are these speed estimates?

These are estimates with real error bars, and any of them could reasonably land anywhere in its published range depending on which runtime you use. The fastest result here: 5,979–15,945 tok/s on B200. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

BERT-Large— 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 1.1 GB, and produces roughly 108 tokens per second. The number of cards able to run it in total: 818.

BERT-Large— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B200, at about 9,965 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: 818.

BERT-Large— how much VRAM does it need?

It needs about 1.1 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.

BERT-Large— 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 1.1 GB and generating roughly 1,856 tokens per second. The fit is comfortable.

BERT-Large— 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 1.1 GB and generating roughly 1,137 tokens per second. The fit is comfortable.

BERT-Large— 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 1.1 GB and generating roughly 1,408 tokens per second. The fit is comfortable.

BERT-Large— 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 1.1 GB and generating roughly 1,669 tokens per second. The fit is comfortable.

BERT-Large— 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.

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