Routing Transformer (WT-103) TPS calculator

Open weights Google Research 79.5M parameters March 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 · 464 tok/s

Fastest card

B200

42,619 tok/s · 180 GB

Which GPUs can run Routing Transformer (WT-103)?

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
42,619 tok/s 25,572–68,191 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	0.8 GB	Q8_0	Comfortable
42,619 tok/s 25,572–68,191 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	0.8 GB	Q8_0	Comfortable
34,033 tok/s 20,420–54,452 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	0.8 GB	Q8_0	Comfortable
34,033 tok/s 20,420–54,452 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	0.8 GB	Q8_0	Comfortable
27,218 tok/s 16,331–43,548 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	0.8 GB	Q8_0	Comfortable
26,051 tok/s 15,631–41,682 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.8 GB	Q8_0	Comfortable
26,051 tok/s 15,631–41,682 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.8 GB	Q8_0	Comfortable
24,932 tok/s 14,959–39,892 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	0.8 GB	Q8_0	Comfortable
22,127 tok/s 13,276–35,404 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	0.8 GB	Q8_0	Comfortable
22,127 tok/s 13,276–35,404 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	0.8 GB	Q8_0	Comfortable
22,127 tok/s 13,276–35,404 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	0.8 GB	Q8_0	Comfortable
20,990 tok/s 12,594–33,584 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	0.8 GB	Q8_0	Comfortable
17,900 tok/s 10,740–28,640 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.8 GB	Q8_0	Comfortable
17,900 tok/s 10,740–28,640 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	0.8 GB	Q8_0	Comfortable
17,900 tok/s 10,740–28,640 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	0.8 GB	Q8_0	Comfortable
17,900 tok/s 10,740–28,640 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.8 GB	Q8_0	Comfortable
17,900 tok/s 10,740–28,640 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	0.8 GB	Q8_0	Comfortable
13,630 tok/s 8,178–21,807 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	0.8 GB	Q8_0	Comfortable
13,630 tok/s 8,178–21,807 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	0.8 GB	Q8_0	Comfortable
11,358 tok/s 6,815–18,173 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	0.8 GB	Q8_0	Comfortable
11,116 tok/s 6,669–17,785 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	0.8 GB	Q8_0	Comfortable
10,868 tok/s 6,521–17,389 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	0.8 GB	Q8_0	Comfortable
10,868 tok/s 6,521–17,389 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	0.8 GB	Q8_0	Comfortable
10,868 tok/s 6,521–17,389 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	0.8 GB	Q8_0	Comfortable
10,868 tok/s 6,521–17,389 · 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: Google Research
Organisation type: Industry
Country: United States of America
Published: 12 March 2020
Authors: Aurko Roy, Mohammad Saffar, Ashish Vaswani, David Grangier

What it does

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

Domain: Language
Task: Language modeling

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: 79.5M
Training data: 103,000,000 tokens

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: SOTA improvement
Record confidence: Confident
Citations: 754
Benchmark data: Routing Transformer

Sources

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

Reference: Efficient Content-Based Sparse Attention with Routing Transformers
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run Routing Transformer (WT-103)

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 Routing Transformer (WT-103)

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

0.8 GB

Fastest

42,619 tok/s

Routing Transformer (WT-103) is small enough at 79.5M parameters that hardware is rarely the obstacle — 818 of the cards we track can run it, including cards several years old.

At the low end, a Tesla C1080 handles it — 4 GB, at Q8_0, for about 464 tokens per second.

At the other end, a B200 generates roughly 42,619 tokens per second on it, on the strength of 8,000 GB/s of memory bandwidth.

About this model

Routing Transformer (WT-103) was published by Google Research, in United States of America, in March 2020. industry is the category the publisher falls under.

It works in Language, and is recorded as doing language modeling.

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

How fast it runs, and why

Half the cards that hold it manage more than 1,196.8 tokens per second, and 818 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.

Training and provenance

The training set ran to roughly 103,000,000 tokens.

The reason it appears in this catalogue at all is sOTA improvement.

Step by step

How to choose a GPU for Routing Transformer (WT-103)

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

Every card here has been checked against Routing Transformer (WT-103) — around 0.8 GB at Q8_0. Capacity is the gate — a card either holds it or it does not.
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 Routing Transformer (WT-103) can slip off a card that handles short questions easily.
03

Decide how much compression you will accept

The quantisation column varies by card, because a bigger card holds a more accurate copy of Routing Transformer (WT-103) — Q8_0 on the smallest card that fits. Set a floor to hold the comparison at one level.
04

Compare tokens per second, not specifications

Sort by speed to see how cards rank for Routing Transformer (WT-103). It will not match a gaming ordering — generation is bound by memory bandwidth, which is why the B200 tops it at 42,619 tok/s.
05

Look at the headroom, not just the fit

The fit column separates cards that just manage Routing Transformer (WT-103) from those with room to spare. Buy for the second if the context might grow.
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 Routing Transformer (WT-103).

Answers

Routing Transformer (WT-103) — common questions

How accurate are these Routing Transformer (WT-103) speed estimates?

They are calculated from specifications rather than measured, and each carries a range — 25,572–68,191 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.

What GPU do I need to run Routing Transformer (WT-103)?

The smallest card in our catalogue that holds Routing Transformer (WT-103) is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 0.8 GB, and produces roughly 464 tokens per second. 818 cards in total can run it.

How fast is Routing Transformer (WT-103) on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 42,619 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 Routing Transformer (WT-103) clear that.

How much VRAM does Routing Transformer (WT-103) 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 Routing Transformer (WT-103) 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 7,938 tokens per second — a comfortable fit.

Can I run Routing Transformer (WT-103) 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 4,861 tokens per second — a comfortable fit.

Can I run Routing Transformer (WT-103) 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 6,020 tokens per second — a comfortable fit.

Can I run Routing Transformer (WT-103) 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 7,139 tokens per second — a comfortable fit.

Is Routing Transformer (WT-103) open source?

Its weights are published, so Routing Transformer (WT-103) 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 Routing Transformer (WT-103) have?

Routing Transformer (WT-103) has 79.5M 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.

Who created Routing Transformer (WT-103)?

Routing Transformer (WT-103) was published by Google Research, based in United States of America, categorised as industry.

When was Routing Transformer (WT-103) released?

Routing Transformer (WT-103) was published in March 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.

What is Routing Transformer (WT-103) used for?

Routing Transformer (WT-103) works in Language, and is recorded as handling language modeling. 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 Routing Transformer (WT-103)?

The weights for Routing Transformer (WT-103) 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 Routing Transformer (WT-103) 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 Routing Transformer (WT-103) is rarely worth using. Every figure here assumes the whole model is on the card.

Would two GPUs run Routing Transformer (WT-103) faster?

Two cards buy memory rather than speed. That matters for Routing Transformer (WT-103) only if one card cannot hold it — 818 can, so a second adds little.

Why does the quantisation differ between cards for Routing Transformer (WT-103)?

A larger card holds a more accurate copy. Across the cards that run Routing Transformer (WT-103), 1 compression levels are used; the floor control above pins it to one.

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