ImageBind TPS calculator

Open weights Meta AI 932M parameters May 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 · 39.6 tok/s

Fastest card

B200

3,635 tok/s · 180 GB

Which GPUs can run ImageBind?

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
3,635 tok/s 2,181–5,817 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	1.7 GB	Q8_0	Comfortable
3,635 tok/s 2,181–5,817 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	1.7 GB	Q8_0	Comfortable
2,903 tok/s 1,742–4,645 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	1.7 GB	Q8_0	Comfortable
2,903 tok/s 1,742–4,645 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	1.7 GB	Q8_0	Comfortable
2,322 tok/s 1,393–3,715 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	1.7 GB	Q8_0	Comfortable
2,222 tok/s 1,333–3,555 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	1.7 GB	Q8_0	Comfortable
2,222 tok/s 1,333–3,555 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	1.7 GB	Q8_0	Comfortable
2,127 tok/s 1,276–3,403 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	1.7 GB	Q8_0	Comfortable
1,887 tok/s 1,132–3,020 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	1.7 GB	Q8_0	Comfortable
1,887 tok/s 1,132–3,020 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	1.7 GB	Q8_0	Comfortable
1,887 tok/s 1,132–3,020 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	1.7 GB	Q8_0	Comfortable
1,790 tok/s 1,074–2,865 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	1.7 GB	Q8_0	Comfortable
1,527 tok/s 916–2,443 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	1.7 GB	Q8_0	Comfortable
1,527 tok/s 916–2,443 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	1.7 GB	Q8_0	Comfortable
1,527 tok/s 916–2,443 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	1.7 GB	Q8_0	Comfortable
1,527 tok/s 916–2,443 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	1.7 GB	Q8_0	Comfortable
1,527 tok/s 916–2,443 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	1.7 GB	Q8_0	Comfortable
1,163 tok/s 698–1,860 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	1.7 GB	Q8_0	Comfortable
1,163 tok/s 698–1,860 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	1.7 GB	Q8_0	Comfortable
969 tok/s 581–1,550 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	1.7 GB	Q8_0	Comfortable
948 tok/s 569–1,517 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	1.7 GB	Q8_0	Comfortable
927 tok/s 556–1,483 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	1.7 GB	Q8_0	Comfortable
927 tok/s 556–1,483 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	1.7 GB	Q8_0	Comfortable
927 tok/s 556–1,483 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	1.7 GB	Q8_0	Comfortable
927 tok/s 556–1,483 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	1.7 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: Meta AI
Organisation type: Industry
Country: United States of America
Published: 9 May 2023
Authors: Rohit Girdhar, Alaaeldin El-Nouby, Zhuang Liu, Mannat Singh, Kalyan Vasudev Alwala, Armand Joulin, Ishan Misra

What it does

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

Domain: Multimodal, Vision, Audio, Language, Image generation, Speech
Task: Image classification, Speech recognition (ASR), Image generation, Language modeling/generation
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: 932M
Training data: tokens
Epochs: 64

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 V100,NVIDIA A100

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: SOTA improvement
Record confidence: Likely
Citations: 1,515

Sources

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

Reference: IMAGEBIND: One Embedding Space To Bind Them All
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run ImageBind

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 ImageBind

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 you need to run it

Minimum card

Tesla C1080

Memory needed

1.7 GB

Fastest

3,635 tok/s

ImageBind is small enough at 932M 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 39.6 tokens per second.

The quickest result comes from a B200 at around 3,635 tokens per second — its 8,000 GB/s of bandwidth is what buys that.

What this model is

ImageBind was published by Meta AI, in United States of America, in May 2023. industry is the category the publisher falls under.

It works in Multimodal, Vision, Audio, Language, Image generation, Speech, and is recorded as doing image classification, Speech recognition (ASR), Image generation, Language modeling/generation.

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

Published weights mean the model runs on your machine rather than someone else's, which is what makes the hardware question below answerable at all.

What decides the speed

Across every card that can run it, the middle of the range is about 102.1 tokens per second, and 806 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.

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.

How it was trained

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

Step by step

How to choose a GPU for ImageBind

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 ImageBind — around 1.7 GB at Q8_0. Capacity is the gate — a card either holds it or it does not.
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 ImageBind can slip off a card that handles short questions easily.
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 ImageBind by squeezing it further than you would want.
04

Sort by speed

The speed ordering for ImageBind is effectively an ordering by memory bandwidth, which is why the B200 tops it at 3,635 tok/s.
05

Look at the headroom, not just the fit

Tight means ImageBind 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

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 ImageBind alone — a card is usually bought for more than one model.

Answers

ImageBind — common questions

Can I run ImageBind on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 1.7 GB and generating roughly 609 tokens per second — a comfortable fit.

Is ImageBind open source?

Its weights are published, so ImageBind 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 ImageBind have?

ImageBind has 932M parameters. used ViT-Huge 630M as an image/video encoder and OpenCLIP-302m as text encoder. 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 ImageBind?

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

When was ImageBind released?

ImageBind was published in May 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.

What is ImageBind used for?

ImageBind works in Multimodal, Vision, Audio, Language, Image generation, Speech, and is recorded as handling image classification, Speech recognition (ASR), Image generation, Language modeling/generation. 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.

Where can I download ImageBind?

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

Can I run ImageBind 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 ImageBind is rarely worth using. Every figure here assumes the whole model is on the card.

Would two GPUs run ImageBind faster?

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

Why does the quantisation differ between cards for ImageBind?

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

How accurate are these ImageBind 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 2,181–5,817 tok/s on the B200 rather than a single number.

What GPU do I need to run ImageBind?

The smallest card in our catalogue that holds ImageBind is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 1.7 GB, and produces roughly 39.6 tokens per second. 818 cards in total can run it.

How fast is ImageBind on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 3,635 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 806 of the cards that can run ImageBind clear that.

How much VRAM does ImageBind need?

About 1.7 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 ImageBind on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q8_0, using about 1.7 GB and generating roughly 677 tokens per second — a comfortable fit.

Can I run ImageBind on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q8_0, using about 1.7 GB and generating roughly 415 tokens per second — a comfortable fit.

Can I run ImageBind on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 1.7 GB and generating roughly 514 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