RoBERTa Base TPS calculator

Open weights Facebook,University of Washington 125M parameters July 2019

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 · 295 tok/s

Fastest card

B200

27,106 tok/s · 180 GB

Which GPUs can run RoBERTa Base?

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
27,106 tok/s 16,264–43,369 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	0.8 GB	Q8_0	Comfortable
27,106 tok/s 16,264–43,369 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	0.8 GB	Q8_0	Comfortable
21,645 tok/s 12,987–34,632 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	0.8 GB	Q8_0	Comfortable
21,645 tok/s 12,987–34,632 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	0.8 GB	Q8_0	Comfortable
17,310 tok/s 10,386–27,697 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	0.8 GB	Q8_0	Comfortable
16,568 tok/s 9,941–26,510 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.8 GB	Q8_0	Comfortable
16,568 tok/s 9,941–26,510 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.8 GB	Q8_0	Comfortable
15,857 tok/s 9,514–25,371 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	0.8 GB	Q8_0	Comfortable
14,073 tok/s 8,444–22,517 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	0.8 GB	Q8_0	Comfortable
14,073 tok/s 8,444–22,517 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	0.8 GB	Q8_0	Comfortable
14,073 tok/s 8,444–22,517 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	0.8 GB	Q8_0	Comfortable
13,350 tok/s 8,010–21,359 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	0.8 GB	Q8_0	Comfortable
11,384 tok/s 6,831–18,215 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.8 GB	Q8_0	Comfortable
11,384 tok/s 6,831–18,215 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	0.8 GB	Q8_0	Comfortable
11,384 tok/s 6,831–18,215 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	0.8 GB	Q8_0	Comfortable
11,384 tok/s 6,831–18,215 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.8 GB	Q8_0	Comfortable
11,384 tok/s 6,831–18,215 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	0.8 GB	Q8_0	Comfortable
8,668 tok/s 5,201–13,870 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	0.8 GB	Q8_0	Comfortable
8,668 tok/s 5,201–13,870 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	0.8 GB	Q8_0	Comfortable
7,224 tok/s 4,334–11,558 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	0.8 GB	Q8_0	Comfortable
7,070 tok/s 4,242–11,311 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	0.8 GB	Q8_0	Comfortable
6,912 tok/s 4,147–11,059 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	0.8 GB	Q8_0	Comfortable
6,912 tok/s 4,147–11,059 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	0.8 GB	Q8_0	Comfortable
6,912 tok/s 4,147–11,059 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	0.8 GB	Q8_0	Comfortable
6,912 tok/s 4,147–11,059 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	0.8 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,University of Washington
Organisation type: Industry,Academia
Country: United States of America
Published: 1 July 2019
Authors: Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov

What it does

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

Domain: Language
Task: Question answering, Language modeling/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: 125M
Training data: tokens
Epochs: 48

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: 1.5 × 10²¹ 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: Open source

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Record confidence: Confident
Citations: 29,641

Sources

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

Reference: RoBERTa: A Robustly Optimized BERT Pretraining Approach
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run RoBERTa Base

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 RoBERTa Base

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

0.8 GB

Fastest

27,106 tok/s

RoBERTa Base is small enough at 125M 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 295 tokens per second.

A B200 is the fastest we calculate for it: about 27,106 tokens per second, from 8,000 GB/s of memory bandwidth.

Where it came from

RoBERTa Base was published by Facebook,University of Washington, in United States of America, in July 2019. The organisation is categorised as industry,Academia.

It works in Language, and is recorded as doing question answering, Language modeling/generation.

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

Understanding the speeds

Across every card that can run it, the middle of the range is about 761.1 tokens per second, and 818 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.

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.

Training and provenance

The training run consumed about 1.5 × 10²¹ FLOP. That figure describes the cost of creating it and has no bearing on how quickly it generates text.

Step by step

How to choose a GPU for RoBERTa Base

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 RoBERTa Base — around 0.8 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 RoBERTa Base 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 RoBERTa Base by squeezing it further than you would want.
04

Rank by throughput rather than spec sheet

Ranking by tokens per second for RoBERTa Base follows memory bandwidth, not core counts, which is why the B200 tops it at 27,106 tok/s.
05

Check the fit verdict before buying

The fit column separates cards that just manage RoBERTa Base from those with room to spare. Buy for the second if the context might grow.
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 RoBERTa Base.

Answers

RoBERTa Base — common questions

What GPU do I need to run RoBERTa Base?

The smallest card in our catalogue that holds RoBERTa Base is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 0.8 GB, and produces roughly 295 tokens per second. 818 cards in total can run it.

How fast is RoBERTa Base on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 27,106 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 818 of the cards that can run RoBERTa Base clear that.

How much VRAM does RoBERTa Base need?

About 0.8 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 RoBERTa Base on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q8_0, using about 0.8 GB and generating roughly 5,048 tokens per second — a comfortable fit.

Can I run RoBERTa Base on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q8_0, using about 0.8 GB and generating roughly 3,091 tokens per second — a comfortable fit.

Can I run RoBERTa Base on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 0.8 GB and generating roughly 3,829 tokens per second — a comfortable fit.

Can I run RoBERTa Base on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 0.8 GB and generating roughly 4,540 tokens per second — a comfortable fit.

Is RoBERTa Base open source?

Its weights are published, so RoBERTa Base 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 RoBERTa Base have?

RoBERTa Base has 125M parameters. "Specifically, we begin by training BERT models with the same configuration as BERTBASE (L = 12, H = 768, A = 12, 110M params" " This adds approximately 15M and 20M additional parameters for BERTBASE and BERTLARGE, respectively" 110M+15M = 125M. 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 RoBERTa Base?

RoBERTa Base was published by Facebook,University of Washington, based in United States of America, categorised as industry,Academia.

When was RoBERTa Base released?

RoBERTa Base was published in July 2019. 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 RoBERTa Base used for?

RoBERTa Base works in Language, and is recorded as handling question answering, 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 RoBERTa Base?

The weights for RoBERTa Base 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 RoBERTa Base?

Around 1.5 × 10²¹ 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 RoBERTa Base if it does not fit in my GPU?

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

Would two GPUs run RoBERTa Base faster?

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

Why does the quantisation differ between cards for RoBERTa Base?

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

How accurate are these RoBERTa Base 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 16,264–43,369 tok/s on the B200 rather than a single number.

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