Retrieval-Augmented Generator TPS calculator

Open weights Facebook,New York University (NYU),University College London (UCL) 626M parameters May 2020

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 · 58.9 tok/s

Fastest card

B200

5,413 tok/s · 180 GB

Which GPUs can run Retrieval-Augmented Generator?

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
5,413 tok/s 3,248–8,660 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	1.4 GB	Q8_0	Comfortable
5,413 tok/s 3,248–8,660 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	1.4 GB	Q8_0	Comfortable
4,322 tok/s 2,593–6,915 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	1.4 GB	Q8_0	Comfortable
4,322 tok/s 2,593–6,915 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	1.4 GB	Q8_0	Comfortable
3,457 tok/s 2,074–5,531 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	1.4 GB	Q8_0	Comfortable
3,308 tok/s 1,985–5,293 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	1.4 GB	Q8_0	Comfortable
3,308 tok/s 1,985–5,293 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	1.4 GB	Q8_0	Comfortable
3,166 tok/s 1,900–5,066 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	1.4 GB	Q8_0	Comfortable
2,810 tok/s 1,686–4,496 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	1.4 GB	Q8_0	Comfortable
2,810 tok/s 1,686–4,496 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	1.4 GB	Q8_0	Comfortable
2,810 tok/s 1,686–4,496 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	1.4 GB	Q8_0	Comfortable
2,666 tok/s 1,599–4,265 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	1.4 GB	Q8_0	Comfortable
2,273 tok/s 1,364–3,637 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	1.4 GB	Q8_0	Comfortable
2,273 tok/s 1,364–3,637 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	1.4 GB	Q8_0	Comfortable
2,273 tok/s 1,364–3,637 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	1.4 GB	Q8_0	Comfortable
2,273 tok/s 1,364–3,637 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	1.4 GB	Q8_0	Comfortable
2,273 tok/s 1,364–3,637 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	1.4 GB	Q8_0	Comfortable
1,731 tok/s 1,039–2,769 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	1.4 GB	Q8_0	Comfortable
1,731 tok/s 1,039–2,769 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	1.4 GB	Q8_0	Comfortable
1,442 tok/s 865–2,308 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	1.4 GB	Q8_0	Comfortable
1,412 tok/s 847–2,259 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	1.4 GB	Q8_0	Comfortable
1,380 tok/s 828–2,208 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	1.4 GB	Q8_0	Comfortable
1,380 tok/s 828–2,208 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	1.4 GB	Q8_0	Comfortable
1,380 tok/s 828–2,208 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	1.4 GB	Q8_0	Comfortable
1,380 tok/s 828–2,208 · 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,New York University (NYU),University College London (UCL)
Organisation type: Industry,Academia,Academia
Country: United States of America, United Kingdom of Great Britain and Northern Ireland
Published: 22 May 2020
Authors: Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, Douwe Kiela

What it does

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

Domain: Language
Task: Question answering, Retrieval-augmented generation
Numerical format: FP16

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: 626M
Training data: 3,074,560 tokens

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 Tesla V100 PCIe 32 GB

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: Highly cited,SOTA improvement
Record confidence: Confident
Citations: 13,907

Sources

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

Reference: Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run Retrieval-Augmented Generator

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 Retrieval-Augmented Generator

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

The hardware side

Minimum card

Tesla C1080

Memory needed

1.4 GB

Fastest

5,413 tok/s

Retrieval-Augmented Generator reaches a parameter count of 626M. 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.

At the low end it is handled by Tesla C1080, with a memory capacity of 4 GB, running it at a compression of Q8_0 and producing around 58.9 tokens per second.

Top of the range is B200, generating roughly 5,413 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

Background

Retrieval-Augmented Generator was published by Facebook,New York University (NYU),University College London (UCL), in the country recorded as United States of America, during May 2020. The publishing organisation is categorised as industry,Academia,Academia.

It works in the domain of Language, and is recorded as performing the task of question answering, Retrieval-augmented generation.

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.

Reading the throughput figures

The median result is around 152.0 tokens per second. Clearing the ten tokens per second that roughly matches reading speed: 809 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.

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

Training consumed a corpus of around 3,074,560 tokens of text.

The reason it appears in this catalogue at all: highly cited,SOTA improvement.

Step by step

How to choose a GPU for Retrieval-Augmented Generator

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

The table lists every card able to hold Retrieval-Augmented Generator, needing around 1.4 GB at a compression of Q8_0. Capacity is the gate — a card either holds it or it does not.
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 a card that seemed fine stops fitting Retrieval-Augmented Generator.
03

Choose how far you will compress it

Each card runs the least-compressed copy it can hold, 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

Sort by speed to see how cards rank for Retrieval-Augmented Generator. It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B200, at 5,413 tok/s.
05

Look at the headroom, not just the fit

The fit column separates cards that just manage it from those with room to spare, in the case of Retrieval-Augmented Generator. 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

See what else that card runs

Following a card through to its own page shows every other model it can hold, which is the question that follows once you have settled on Retrieval-Augmented Generator.

Answers

Retrieval-Augmented Generator — common questions

Retrieval-Augmented Generator— 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.

Retrieval-Augmented Generator— can I run it if it does not fit in my GPU?

It can be split between the card and system memory, but it generates painfully slowly that way. Every figure here assumes the whole model is resident on the card.

Retrieval-Augmented Generator— would two GPUs run it faster?

Two cards buy memory rather than speed, which matters only if one card cannot hold it. The number that can: 818. So a second card is rarely the answer here.

Retrieval-Augmented Generator— 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.

Retrieval-Augmented Generator— how accurate are these speed estimates?

Every figure is derived from memory bandwidth and model size, not benchmarked, which is why each is published as a range rather than a single number. One example: 3,248–8,660 tok/s on B200. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

Retrieval-Augmented Generator— 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 1.4 GB, and produces roughly 58.9 tokens per second. The number of cards able to run it in total: 818.

Retrieval-Augmented Generator— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B200, at about 5,413 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: 809.

Retrieval-Augmented Generator— how much VRAM does it need?

It needs about 1.4 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.

Retrieval-Augmented Generator— 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 1.4 GB and generating roughly 1,008 tokens per second. The fit is comfortable.

Retrieval-Augmented Generator— 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 1.4 GB and generating roughly 617 tokens per second. The fit is comfortable.

Retrieval-Augmented Generator— 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 1.4 GB and generating roughly 765 tokens per second. The fit is comfortable.

Retrieval-Augmented Generator— 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 1.4 GB and generating roughly 907 tokens per second. The fit is comfortable.

Retrieval-Augmented Generator— 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.

Retrieval-Augmented Generator— how many parameters does it have?

It has a parameter count of 626M. "Our RAG models contain the trainable parameters for the BERT-base query and document encoder of DPR, with 110M parameters each (although we do not train the document encoder ourselves) and 406M trainable parameters from BART-large, 406M parameters, making a total of 626M trainable 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.

Retrieval-Augmented Generator— who created it?

It was published by Facebook,New York University (NYU),University College London (UCL), based in United States of America, an organisation categorised as industry,Academia,Academia.

Retrieval-Augmented Generator— when was it released?

It was published in May 2020. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

Retrieval-Augmented Generator— what is it used for?

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

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