Masked Autoencoders ViT-H TPS calculator

Open weights Facebook AI Research 632M parameters November 2021

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 · 58.3 tok/s

Fastest card

B200

5,361 tok/s · 180 GB

Which GPUs can run Masked Autoencoders ViT-H?

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
5,361 tok/s 3,217–8,578 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	1.4 GB	Q8_0	Comfortable
5,361 tok/s 3,217–8,578 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	1.4 GB	Q8_0	Comfortable
4,281 tok/s 2,569–6,850 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	1.4 GB	Q8_0	Comfortable
4,281 tok/s 2,569–6,850 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	1.4 GB	Q8_0	Comfortable
3,424 tok/s 2,054–5,478 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	1.4 GB	Q8_0	Comfortable
3,277 tok/s 1,966–5,243 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	1.4 GB	Q8_0	Comfortable
3,277 tok/s 1,966–5,243 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	1.4 GB	Q8_0	Comfortable
3,136 tok/s 1,882–5,018 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	1.4 GB	Q8_0	Comfortable
2,783 tok/s 1,670–4,453 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	1.4 GB	Q8_0	Comfortable
2,783 tok/s 1,670–4,453 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	1.4 GB	Q8_0	Comfortable
2,783 tok/s 1,670–4,453 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	1.4 GB	Q8_0	Comfortable
2,640 tok/s 1,584–4,225 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	1.4 GB	Q8_0	Comfortable
2,252 tok/s 1,351–3,603 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	1.4 GB	Q8_0	Comfortable
2,252 tok/s 1,351–3,603 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	1.4 GB	Q8_0	Comfortable
2,252 tok/s 1,351–3,603 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	1.4 GB	Q8_0	Comfortable
2,252 tok/s 1,351–3,603 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	1.4 GB	Q8_0	Comfortable
2,252 tok/s 1,351–3,603 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	1.4 GB	Q8_0	Comfortable
1,714 tok/s 1,029–2,743 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	1.4 GB	Q8_0	Comfortable
1,714 tok/s 1,029–2,743 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	1.4 GB	Q8_0	Comfortable
1,429 tok/s 857–2,286 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	1.4 GB	Q8_0	Comfortable
1,398 tok/s 839–2,237 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	1.4 GB	Q8_0	Comfortable
1,367 tok/s 820–2,187 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	1.4 GB	Q8_0	Comfortable
1,367 tok/s 820–2,187 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	1.4 GB	Q8_0	Comfortable
1,367 tok/s 820–2,187 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	1.4 GB	Q8_0	Comfortable
1,367 tok/s 820–2,187 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	1.4 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: Facebook AI Research
Organisation type: Industry
Country: United States of America, France
Published: 11 November 2021
Authors: Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick

What it does

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

Domain: Vision
Task: Semantic segmentation, Image classification, Image generation
Approach: Self-supervised learning
Base model: ViT-Huge/14

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: 632M
Training data: 327,978,752 tokens
Epochs: 1,600

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: 4.6 × 10²⁰ FLOP
How it was established: Hardware,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.

Wall-clock time: 69 hours

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: Open (non-commercial)

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,SOTA improvement
Record confidence: Speculative
Citations: 11,449

Sources

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

Reference: Masked Autoencoders Are Scalable Vision Learners
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run Masked Autoencoders ViT-H

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 Masked Autoencoders ViT-H

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

1.4 GB

Fastest

5,361 tok/s

Masked Autoencoders ViT-H is small enough at 632M 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 Q8_0 and producing around 58.3 tokens per second.

The quickest result comes from a B200 at around 5,361 tokens per second — its 8,000 GB/s of bandwidth is what buys that.

Where it came from

Masked Autoencoders ViT-H was published by Facebook AI Research, in United States of America, in November 2021. industry is the category the publisher falls under.

It works in Vision, and is recorded as doing semantic segmentation, Image classification, Image generation.

It is derived from ViT-Huge/14 rather than trained from scratch, which is the usual way a specialised model is produced.

Because its weights were released, nothing about running it depends on a provider staying available — it is yours once downloaded.

Understanding the speeds

The median result is around 150.5 tokens per second; 809 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.

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

Producing it required around 4.6 × 10²⁰ FLOP of arithmetic, which is a statement about the training budget rather than about inference.

Around 327,978,752 tokens went into training it.

Its inclusion criterion is highly cited,SOTA improvement.

Step by step

How to choose a GPU for Masked Autoencoders ViT-H

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 Masked Autoencoders ViT-H actually needs — around 1.4 GB at Q8_0. No amount of processing power compensates for a card that cannot hold it.
02

Decide how long your conversations run

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

Choose how far you will compress it

Each card runs the least-compressed copy it can hold — Q8_0 on the smallest card that fits. Setting a floor drops the cards that only manage Masked Autoencoders ViT-H by squeezing it further than you would want.
04

Rank by throughput rather than spec sheet

Sort by speed to see how cards rank for Masked Autoencoders ViT-H. It will not match a gaming ordering — generation is bound by memory bandwidth, which is why the B200 tops it at 5,361 tok/s.
05

Check the fit verdict before buying

A tight fit runs Masked Autoencoders ViT-H but leaves nothing spare for a longer conversation; comfortable has headroom. If you expect to grow the context, buy for comfortable.
06

See what else that card runs

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

Answers

Masked Autoencoders ViT-H — common questions

Where can I download Masked Autoencoders ViT-H?

The weights for Masked Autoencoders ViT-H 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 Masked Autoencoders ViT-H?

Around 4.6 × 10²⁰ 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 Masked Autoencoders ViT-H if it does not fit in my GPU?

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

Would two GPUs run Masked Autoencoders ViT-H faster?

Two cards buy memory rather than speed. That matters for Masked Autoencoders ViT-H only if one card cannot hold it — 818 can, so a second adds little.

Why does the quantisation differ between cards for Masked Autoencoders ViT-H?

Each card is shown running the least-compressed copy it can hold, and Masked Autoencoders ViT-H appears at 1 different compression levels across the cards that fit it. Bigger cards get the more accurate version.

How accurate are these Masked Autoencoders ViT-H speed estimates?

Every figure is derived from memory bandwidth and model size, not benchmarked. That is why each is published as a range such as 3,217–8,578 tok/s on the B200 rather than a single number.

What GPU do I need to run Masked Autoencoders ViT-H?

The smallest card in our catalogue that holds Masked Autoencoders ViT-H is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 1.4 GB, and produces roughly 58.3 tokens per second. 818 cards in total can run it.

How fast is Masked Autoencoders ViT-H on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 5,361 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 809 of the cards that can run Masked Autoencoders ViT-H clear that.

How much VRAM does Masked Autoencoders ViT-H need?

About 1.4 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 Masked Autoencoders ViT-H on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q8_0, using about 1.4 GB and generating roughly 999 tokens per second — a comfortable fit.

Can I run Masked Autoencoders ViT-H on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q8_0, using about 1.4 GB and generating roughly 611 tokens per second — a comfortable fit.

Can I run Masked Autoencoders ViT-H on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 1.4 GB and generating roughly 757 tokens per second — a comfortable fit.

Can I run Masked Autoencoders ViT-H on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 1.4 GB and generating roughly 898 tokens per second — a comfortable fit.

Is Masked Autoencoders ViT-H open source?

Its weights are published, so Masked Autoencoders ViT-H 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 Masked Autoencoders ViT-H have?

Masked Autoencoders ViT-H has 632M parameters. Three models: ViT-B (86M), ViT-L (304M), ViT-H (632M). 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 Masked Autoencoders ViT-H?

Masked Autoencoders ViT-H was published by Facebook AI Research, based in United States of America, categorised as industry.

When was Masked Autoencoders ViT-H released?

Masked Autoencoders ViT-H was published in November 2021. 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 Masked Autoencoders ViT-H used for?

Masked Autoencoders ViT-H works in Vision, and is recorded as handling semantic segmentation, Image classification, Image generation. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

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