Segment Anything Model TPS calculator

Open weights Meta AI 636M parameters April 2023

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 · 58.0 tok/s

Fastest card

B200

5,327 tok/s · 180 GB

Which GPUs can run Segment Anything Model?

Set the inputs, read the answer

A longer conversation needs more memory, which can push this model off smaller cards.

Hides cards that would only fit the model by compressing it below this point.

818 cards match

Calculating
Needs Quantisation Fit
5,327 tok/s

3,196–8,524 · low confidence

B200 NVIDIA 180 GB 8,000 GB/s Jan 2024 1.4 GB Q8_0 Comfortable
5,327 tok/s

3,196–8,524 · low confidence

B300 NVIDIA 288 GB 8,000 GB/s Sep 2025 1.4 GB Q8_0 Comfortable
4,254 tok/s

2,552–6,807 · low confidence

Radeon Instinct MI350X AMD 288 GB 8,190 GB/s Jan 2025 1.4 GB Q8_0 Comfortable
4,254 tok/s

2,552–6,807 · low confidence

Radeon Instinct MI355X AMD 288 GB 8,190 GB/s Jan 2025 1.4 GB Q8_0 Comfortable
3,402 tok/s

2,041–5,444 · low confidence

Radeon Instinct MI300 AMD 128 GB 6,550 GB/s Jan 2023 1.4 GB Q8_0 Comfortable
3,256 tok/s

1,954–5,210 · low confidence

H200 NVL NVIDIA 141 GB 4,890 GB/s Nov 2024 1.4 GB Q8_0 Comfortable
3,256 tok/s

1,954–5,210 · low confidence

H200 SXM 141 GB NVIDIA 141 GB 4,890 GB/s Nov 2024 1.4 GB Q8_0 Comfortable
3,117 tok/s

1,870–4,986 · low confidence

Radeon Instinct MI325X AMD 256 GB 6,000 GB/s Oct 2024 1.4 GB Q8_0 Comfortable
2,766 tok/s

1,660–4,425 · low confidence

Radeon Instinct MI300A AMD 128 GB 5,325 GB/s Dec 2023 1.4 GB Q8_0 Comfortable
2,766 tok/s

1,660–4,425 · low confidence

Radeon Instinct MI300X AMD 192 GB 5,325 GB/s Dec 2023 1.4 GB Q8_0 Comfortable
2,766 tok/s

1,660–4,425 · low confidence

Radeon Instinct MI308X AMD 192 GB 5,325 GB/s Dec 2023 1.4 GB Q8_0 Comfortable
2,624 tok/s

1,574–4,198 · low confidence

H100 NVL 94 GB NVIDIA 94 GB 3,940 GB/s Mar 2023 1.4 GB Q8_0 Comfortable
2,238 tok/s

1,343–3,580 · low confidence

H100 PCIe 96 GB NVIDIA 96 GB 3,360 GB/s Mar 2023 1.4 GB Q8_0 Comfortable
2,238 tok/s

1,343–3,580 · low confidence

H100 SXM5 80 GB NVIDIA 80 GB 3,360 GB/s Oct 2022 1.4 GB Q8_0 Comfortable
2,238 tok/s

1,343–3,580 · low confidence

H100 SXM5 94 GB NVIDIA 94 GB 3,360 GB/s Mar 2023 1.4 GB Q8_0 Comfortable
2,238 tok/s

1,343–3,580 · low confidence

H100 SXM5 96 GB NVIDIA 96 GB 3,360 GB/s Mar 2023 1.4 GB Q8_0 Comfortable
2,238 tok/s

1,343–3,580 · low confidence

H800 SXM5 NVIDIA 80 GB 3,360 GB/s Mar 2023 1.4 GB Q8_0 Comfortable
1,704 tok/s

1,022–2,726 · low confidence

Radeon Instinct MI250 AMD 128 GB 3,280 GB/s Nov 2021 1.4 GB Q8_0 Comfortable
1,704 tok/s

1,022–2,726 · low confidence

Radeon Instinct MI250X AMD 128 GB 3,280 GB/s Nov 2021 1.4 GB Q8_0 Comfortable
1,420 tok/s

852–2,272 · low confidence

Data Center GPU Max 1550 Intel 128 GB 3,280 GB/s Jan 2023 1.4 GB Q8_0 Comfortable
1,389 tok/s

834–2,223 · low confidence

Data Center GPU Max Subsystem Intel 128 GB 3,210 GB/s Jan 2023 1.4 GB Q8_0 Comfortable
1,358 tok/s

815–2,174 · low confidence

A100 SXM4 80 GB NVIDIA 80 GB 2,040 GB/s Nov 2020 1.4 GB Q8_0 Comfortable
1,358 tok/s

815–2,174 · low confidence

A100X NVIDIA 80 GB 2,040 GB/s Jun 2021 1.4 GB Q8_0 Comfortable
1,358 tok/s

815–2,174 · low confidence

A800 SXM4 80 GB NVIDIA 80 GB 2,040 GB/s Aug 2022 1.4 GB Q8_0 Comfortable
1,358 tok/s

815–2,174 · 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
Meta AI
Organisation type
Industry
Country
United States of America
Published
5 April 2023
Authors
Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C. Berg, Wan-Yen Lo, 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
Image segmentation
Approach
Supervised
Base model
ViT-Huge/14
Numerical format
FP32

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
636M

From Facebook website: https://segment-anything.com/ "How big is the model? The image encoder has 632M parameters. The prompt encoder and mask decoder have 4M parameters."

Training data
1,100,000,000 tokens

"SA-1B contains 11M diverse, high-resolution, licensed, and privacy protecting images and 1.1B high-quality segmentation masks." segmentation mask is a map that identifies segments in an image

Epochs
2

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
7.8 × 10²¹ FLOP

"SAM was trained on 256 A100 GPUS for 68 hours. We acknowledge the environmental impact and cost of training large scale models. The environmental impact of training the released SAM model is approximately 6963 kWh" 68*256 A100-hours = 17408 hours * 3600 * 312 trillion * 0.4 (utilization assumption for image models) = 7.82e21 max A100 power is 400W. 6,963,000 watt-hours / 400 watts = 17407.5 hours (so they probably just calculated backwards from power rating, and this doesn't give any info on…

How it was established
Hardware
Fine-tuning compute
7.8 × 10²¹ FLOP

see Training Compute notes

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
Chips used
256
Wall-clock time
68 hours

"SAM was trained on 256 A100 GPUS for 68 hours"

Power draw
204.1 kW
Compute cost
$15,888

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

Apache 2.0 license don't see pretrain code in the repo, could be wrong https://github.com/facebookresearch/segment-anything

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,Historical significance
Record confidence
Confident
Citations
13,486

Sources

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

Reference
Segment Anything
Last updated
25 May 2026

The extremes

The ten fastest GPUs that run Segment Anything Model

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.

  1. 01 B300 288 GB · 8,000 GB/s · Q8_0 5,327 tok/s
  2. 02 B200 180 GB · 8,000 GB/s · Q8_0 5,327 tok/s
  3. 03 Radeon Instinct MI350X 288 GB · 8,190 GB/s · Q8_0 4,254 tok/s
  4. 04 Radeon Instinct MI355X 288 GB · 8,190 GB/s · Q8_0 4,254 tok/s
  5. 05 Radeon Instinct MI300 128 GB · 6,550 GB/s · Q8_0 3,402 tok/s
  6. 06 H200 NVL 141 GB · 4,890 GB/s · Q8_0 3,256 tok/s
  7. 07 H200 SXM 141 GB 141 GB · 4,890 GB/s · Q8_0 3,256 tok/s
  8. 08 Radeon Instinct MI325X 256 GB · 6,000 GB/s · Q8_0 3,117 tok/s
  9. 09 Radeon Instinct MI300A 128 GB · 5,325 GB/s · Q8_0 2,766 tok/s
  10. 10 Radeon Instinct MI300X 192 GB · 5,325 GB/s · Q8_0 2,766 tok/s

The smallest GPUs that still run Segment Anything Model

The cheapest route in, by memory capacity. A tight fit runs the model but leaves nothing spare for a longer conversation.

  1. 01 GeForce RTX 4010 4 GB · needs 1.4 GB · Q8_0 · comfortable 63.9 tok/s
  2. 02 RTX A400 4 GB · needs 1.4 GB · Q8_0 · comfortable 63.9 tok/s
  3. 03 RTX 500 Mobile Ada Generation 4 GB · needs 1.4 GB · Q8_0 · comfortable 85.2 tok/s
  4. 04 GeForce RTX 3050 A Mobile 4 GB · needs 1.4 GB · Q8_0 · comfortable 128 tok/s
  5. 05 Jetson Orin Nano 4 GB 4 GB · needs 1.4 GB · Q8_0 · comfortable 22.7 tok/s
  6. 06 Radeon RX 6450M 4 GB · needs 1.4 GB · Q8_0 · comfortable 66.5 tok/s
  7. 07 Radeon RX 6550M 4 GB · needs 1.4 GB · Q8_0 · comfortable 74.8 tok/s
  8. 08 Radeon RX 6550S 4 GB · needs 1.4 GB · Q8_0 · comfortable 66.5 tok/s
  9. 09 Arc A310 4 GB · needs 1.4 GB · Q8_0 · comfortable 53.7 tok/s
  10. 10 Arc Pro A30M 4 GB · needs 1.4 GB · Q8_0 · comfortable 55.4 tok/s

What the numbers mean

What you need to run it

Minimum card

Tesla C1080

Memory needed

1.4 GB

Fastest

5,327 tok/s

Segment Anything Model is small enough at 636M parameters that hardware is rarely the obstacle — 818 of the cards we track can run it, including cards several years old.

The least hardware that works is a Tesla C1080. Its 4 GB is enough at Q8_0 compression, giving roughly 58.0 tokens per second.

Top of the range is the B200, at roughly 5,327 tokens per second thanks to 8,000 GB/s of bandwidth.

Where it came from

Segment Anything Model was published by Meta AI, in United States of America, in April 2023. The organisation is categorised as industry.

It works in Vision, and is recorded as doing image segmentation.

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

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.

Understanding the speeds

Half the cards that hold it manage more than 149.6 tokens per second, and 809 exceed reading speed outright.

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.

Memory here is estimated from size rather than computed from the architecture, which is not recorded for this model — the numbers are indicative rather than exact.

How it was trained

The training run consumed about 7.8 × 10²¹ FLOP, on NVIDIA A100. That figure describes the cost of creating it and has no bearing on how quickly it generates text.

The training set ran to roughly 1,100,000,000 tokens.

It is tracked in the underlying dataset for one reason in particular: highly cited,Historical significance.

Step by step

How to choose a GPU for Segment Anything Model

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

  1. 01

    Start from the memory column

    Every card here has been checked against Segment Anything Model — around 1.4 GB at Q8_0. Capacity is the gate — a card either holds it or it does not.

  2. 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: at long context Segment Anything Model can slip off a card that handles short questions easily.

  3. 03

    Set a quality floor

    The quantisation column varies by card, because a bigger card holds a more accurate copy of Segment Anything Model — Q8_0 on the smallest card that fits. Set a floor to hold the comparison at one level.

  4. 04

    Sort by speed

    Sort by speed to see how cards rank for Segment Anything Model. It will not match a gaming ordering — generation is bound by memory bandwidth, which is why the B200 tops it at 5,327 tok/s.

  5. 05

    Read the fit column last

    Tight means Segment Anything Model 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.

  6. 06

    Open the card you have settled on

    Each card page repeats this sweep for every model we hold. It answers what else the hardware is good for, beyond Segment Anything Model.

Answers

Segment Anything Model — common questions

01

Can I run Segment Anything Model 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 608 tokens per second — a comfortable fit.

02

Can I run Segment Anything Model 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 753 tokens per second — a comfortable fit.

03

Can I run Segment Anything Model 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 892 tokens per second — a comfortable fit.

04

Is Segment Anything Model open source?

Its weights are published, so Segment Anything Model 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.

05

How many parameters does Segment Anything Model have?

Segment Anything Model has 636M parameters. From Facebook website: https://segment-anything.com/ "How big is the model? The image encoder has 632M parameters. The prompt encoder and mask decoder have 4M parameters.". 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.

06

Who created Segment Anything Model?

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

07

When was Segment Anything Model released?

Segment Anything Model was published in April 2023. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

08

What is Segment Anything Model used for?

Segment Anything Model works in Vision, and is recorded as handling image segmentation. A model can carry several of each, so these are the areas it was built for rather than a limit on what it will attempt.

09

Where can I download Segment Anything Model?

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

10

How much compute was used to train Segment Anything Model?

Around 7.8 × 10²¹ FLOP, on NVIDIA A100. 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.

11

Can I run Segment Anything Model if it does not fit in my GPU?

Only by offloading, which is usually a false economy: the part in system memory drags the whole thing down. Our figures for Segment Anything Model assume it is fully resident.

12

Would two GPUs run Segment Anything Model faster?

Capacity adds across cards; throughput does not. Since 818 of the cards we track already hold Segment Anything Model on their own, a second card is rarely the answer here.

13

Why does the quantisation differ between cards for Segment Anything Model?

Because capacity varies, so does how hard Segment Anything Model has to be squeezed — 1 distinct levels appear in the table above. Set a minimum quality to compare at one.

14

How accurate are these Segment Anything Model speed estimates?

They are calculated from specifications rather than measured, and each carries a range — 3,196–8,524 tok/s on the B200, for instance. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

15

What GPU do I need to run Segment Anything Model?

The smallest card in our catalogue that holds Segment Anything Model is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 1.4 GB, and produces roughly 58.0 tokens per second. 818 cards in total can run it.

16

How fast is Segment Anything Model on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 5,327 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 Segment Anything Model clear that.

17

How much VRAM does Segment Anything Model 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.

18

Can I run Segment Anything Model 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 992 tokens per second — a comfortable fit.

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