BEIT-3 TPS calculator

Open weights Microsoft 1.9B parameters August 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

818 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Tesla C1080

4 GB · Q8_0 · 19.4 tok/s

Fastest card

B200

1,783 tok/s · 180 GB

Which GPUs can run BEIT-3?

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
1,783 tok/s 1,070–2,853 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	2.7 GB	Q8_0	Comfortable
1,783 tok/s 1,070–2,853 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	2.7 GB	Q8_0	Comfortable
1,424 tok/s 854–2,278 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	2.7 GB	Q8_0	Comfortable
1,424 tok/s 854–2,278 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	2.7 GB	Q8_0	Comfortable
1,139 tok/s 683–1,822 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	2.7 GB	Q8_0	Comfortable
1,090 tok/s 654–1,744 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	2.7 GB	Q8_0	Comfortable
1,090 tok/s 654–1,744 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	2.7 GB	Q8_0	Comfortable
1,043 tok/s 626–1,669 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	2.7 GB	Q8_0	Comfortable
926 tok/s 556–1,481 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	2.7 GB	Q8_0	Comfortable
926 tok/s 556–1,481 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	2.7 GB	Q8_0	Comfortable
926 tok/s 556–1,481 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	2.7 GB	Q8_0	Comfortable
878 tok/s 527–1,405 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	2.7 GB	Q8_0	Comfortable
749 tok/s 449–1,198 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	2.7 GB	Q8_0	Comfortable
749 tok/s 449–1,198 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	2.7 GB	Q8_0	Comfortable
749 tok/s 449–1,198 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	2.7 GB	Q8_0	Comfortable
749 tok/s 449–1,198 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	2.7 GB	Q8_0	Comfortable
749 tok/s 449–1,198 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	2.7 GB	Q8_0	Comfortable
570 tok/s 342–912 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	2.7 GB	Q8_0	Comfortable
570 tok/s 342–912 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	2.7 GB	Q8_0	Comfortable
475 tok/s 285–760 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	2.7 GB	Q8_0	Comfortable
465 tok/s 279–744 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	2.7 GB	Q8_0	Comfortable
455 tok/s 273–728 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	2.7 GB	Q8_0	Comfortable
455 tok/s 273–728 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	2.7 GB	Q8_0	Comfortable
455 tok/s 273–728 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	2.7 GB	Q8_0	Comfortable
455 tok/s 273–728 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	2.7 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: Microsoft
Organisation type: Industry
Country: United States of America
Published: 22 August 2022
Authors: Image as a Foreign Language: BEiT Pretraining for All Vision and Vision-Language Tasks

What it does

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

Domain: Multimodal, Vision, Language
Task: Object detection, Semantic segmentation, Image classification, Visual question answering, Image captioning, Language generation

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: 1.9B
Training data: 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: 7 × 10¹⁹ FLOP
How it was established: Operation counting

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: Unreleased

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: SOTA improvement
Record confidence: Likely
Citations: 721

Sources

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

Reference: Image as a Foreign Language: BEiT Pretraining for All Vision and Vision-Language Tasks
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run BEIT-3

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 BEIT-3

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

Tesla C1080

Memory needed

2.7 GB

Fastest

1,783 tok/s

BEIT-3 is small enough at 1.9B parameters that hardware is rarely the obstacle — 818 of the cards we track can run it, including cards several years old.

The entry point is the Tesla C1080: 4 GB of memory, Q8_0 compression, roughly 19.4 tokens per second.

At the other end, a B200 generates roughly 1,783 tokens per second on it, on the strength of 8,000 GB/s of memory bandwidth.

Background

BEIT-3 was published by Microsoft, in United States of America, in August 2022. It comes out of industry.

It works in Multimodal, Vision, Language, and is recorded as doing object detection, Semantic segmentation, Image classification, Visual question answering, Image captioning, Language generation.

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

Reading the throughput figures

Across every card that can run it, the middle of the range is about 50.1 tokens per second, and 789 of them clear the ten tokens per second that roughly matches reading speed.

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 internal architecture is not on file, so memory is approximated from the parameter count and marked accordingly. Expect the real figure to differ, more so at long context.

What went into building it

Training it took roughly 7 × 10¹⁹ FLOP of computation — a measure of what producing the model cost, not of how fast it answers.

It is tracked in the underlying dataset for one reason in particular: sOTA improvement.

Step by step

How to choose a GPU for BEIT-3

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

The table lists every card that can hold BEIT-3 — around 2.7 GB at Q8_0. That figure, not the card's headline performance, is what decides whether it runs.
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: at long context BEIT-3 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 BEIT-3 — Q8_0 on the smallest card that fits. Set a floor to hold the comparison at one level.
04

Sort by speed

Sort by speed to see how cards rank for BEIT-3. It will not match a gaming ordering — generation is bound by memory bandwidth, which is why the B200 tops it at 1,783 tok/s.
05

Read the fit column last

Tight means BEIT-3 loads and works, with no room to raise the context later. Comfortable means you can. The difference matters more than a few tokens per second.
06

See what else that card runs

Each card page repeats this sweep for every model we hold. It answers what else the hardware is good for, beyond BEIT-3.

Answers

BEIT-3 — common questions

How accurate are these BEIT-3 speed estimates?

These are estimates with real error bars. The fastest result here, 1,070–2,853 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 BEIT-3?

The smallest card in our catalogue that holds BEIT-3 is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 2.7 GB, and produces roughly 19.4 tokens per second. 818 cards in total can run it.

How fast is BEIT-3 on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 1,783 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 789 of the cards that can run BEIT-3 clear that.

How much VRAM does BEIT-3 need?

About 2.7 GB at Q8_0 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 BEIT-3 on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q8_0, using about 2.7 GB and generating roughly 332 tokens per second — a comfortable fit.

Can I run BEIT-3 on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q8_0, using about 2.7 GB and generating roughly 203 tokens per second — a comfortable fit.

Can I run BEIT-3 on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 2.7 GB and generating roughly 252 tokens per second — a comfortable fit.

Can I run BEIT-3 on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 2.7 GB and generating roughly 299 tokens per second — a comfortable fit.

Is BEIT-3 open source?

Its weights are published, so BEIT-3 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 BEIT-3 have?

BEIT-3 has 1.9B parameters. 1.9B from Table 2. 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 BEIT-3?

BEIT-3 was published by Microsoft, based in United States of America, categorised as industry.

When was BEIT-3 released?

BEIT-3 was published in August 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 BEIT-3 used for?

BEIT-3 works in Multimodal, Vision, Language, and is recorded as handling object detection, Semantic segmentation, Image classification, Visual question answering, Image captioning, Language generation. These are the areas it was designed around; they describe intent rather than a hard boundary.

Where can I download BEIT-3?

The weights for BEIT-3 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 BEIT-3?

Around 7 × 10¹⁹ FLOP. 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.

Can I run BEIT-3 if it does not fit in my GPU?

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

Would two GPUs run BEIT-3 faster?

A second card roughly doubles the memory available but not the generation rate. With 818 cards already able to run BEIT-3 alone, the case for pairing is weak.

Why does the quantisation differ between cards for BEIT-3?

A larger card holds a more accurate copy. Across the cards that run BEIT-3, 1 compression levels are used; the floor control above pins it to one.

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