Switch TPS calculator

Open weights Google 1.6T parameters January 2021

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

0 of 818 cards that can run it

Which GPUs can run Switch?

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.

0 cards match

Calculating
Needs Quantisation Fit

No card in our catalogue can run this model with these settings.

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
Organisation type
Industry
Country
United States of America
Published
11 January 2021
Authors
William Fedus, Barret Zoph, Noam Shazeer

What it does

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

Domain
Language
Task
Text autocompletion
Approach
Self-supervised learning
Numerical format
BF16

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
1.6T

"Combining expert, model and data parallelism, we design two large Switch Transformer models, one with 395 billion and 1.6 trillion parameters" Table 9 gives more precise count of 1571B parameters

Training data
86,400,000,000 tokens

"In our protocol we pre-train with 2^20 (1,048,576) tokens per batch for 550k steps amounting to 576B total tokens." 1 token ~ 0.75 words

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
8.2 × 10²² FLOP

Table 4 https://arxiv.org/ftp/arxiv/papers/2104/2104.10350.pdf

How it was established
Third-party estimation

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
Google TPU v3
Chips used
1,024
Chip-hours
663,552
Wall-clock time
648 hours (27 days)

see table 4 in https://arxiv.org/ftp/arxiv/papers/2104/2104.10350.pdf

Hardware utilisation
HFU 28.0%

Table 4 in https://arxiv.org/pdf/2104.10350 gives measured performance of 34.4 TFLOP/s, vs. peak achievable FLOP/s of 123 TFLOP/s on the TPUv3 being used. HFU = 34.4/123 = 0.27967

Power draw
935.4 kW
Compute cost
$145,101

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 for weights: https://huggingface.co/google/switch-c-2048 paper links to this repo but not clear that the training hyperparams for Switch are here: https://github.com/google-research/t5x

How it is classified

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

Frontier model
Yes
Why it is tracked
Highly cited,SOTA improvement

" On ANLI (Nie et al., 2019), Switch XXL improves over the prior state-of-the-art to get a 65.7 accuracy versus the prior best of 49.4 (Yang et al., 2020)... Finally, we also conduct an early examination of the model’s knowledge with three closed-book knowledge-based tasks: Natural Questions, WebQuestions and TriviaQA, without additional pre-training using Salient Span Masking (Guu et al., 2020). In all three cases, we observe improvements over the prior stateof-the-art T5-XXL model (without SSM…

Record confidence
Confident
Citations
3,888

Sources

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

Reference
Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity
Last updated
25 May 2026

What the numbers mean

What you need to run it

At 1.6T parameters, Switch is beyond what any single graphics card holds. Running it means either splitting it across several cards or renting hardware built for the job — 0 of the cards we track can hold it on their own, and all of them are datacentre parts.

Background

Switch was published by Google, in United States of America, in January 2021. The organisation is categorised as industry.

It works in Language, and is recorded as doing text autocompletion.

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

Training and provenance

The training run consumed about 8.2 × 10²² FLOP, on Google TPU v3. That figure describes the cost of creating it and has no bearing on how quickly it generates text.

Around 86,400,000,000 tokens went into training it.

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

Step by step

How to choose a GPU for Switch

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

  1. 01

    Read the memory figure first

    Look at what Switch actually needs. No amount of processing power compensates for a card that cannot hold it.

  2. 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 Switch can slip off a card that handles short questions easily.

  3. 03

    Decide how much compression you will accept

    Compression is what makes Switch fit smaller cards, at some cost in accuracy. A minimum quality removes the ones that go too far.

  4. 04

    Sort by speed

    Sort by speed to see how cards rank for Switch. It will not match a gaming ordering — generation is bound by memory bandwidth.

  5. 05

    Read the fit column last

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

  6. 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 Switch is settled.

Answers

Switch — common questions

01

What is Switch used for?

Switch works in Language, and is recorded as handling text autocompletion. These are the areas it was designed around; they describe intent rather than a hard boundary.

02

Where can I download Switch?

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

03

How much compute was used to train Switch?

Around 8.2 × 10²² FLOP, on Google TPU v3. 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.

04

Can I run Switch if it does not fit in my GPU?

It can be split between the card and system memory, but Switch generates painfully slowly that way — the nearest miss we calculate is short by 691.9 GB. Nothing on this page assumes offloading.

05

Would two GPUs run Switch faster?

Two cards buy memory rather than speed. That matters for Switch only if one card cannot hold it — 0 can, so a second adds little.

06

Why does the quantisation differ between cards for Switch?

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

07

How accurate are these Switch speed estimates?

These are estimates with real error bars. The fastest result here, the range beneath each figure, could reasonably land anywhere in its published range depending on which runtime you use.

08

Is Switch open source?

Its weights are published, so Switch 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.

09

How many parameters does Switch have?

Switch has 1.6T parameters. "Combining expert, model and data parallelism, we design two large Switch Transformer models, one with 395 billion and 1.6 trillion parameters" Table 9 gives more precise count of 1571B 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.

10

Who created Switch?

Switch was published by Google, based in United States of America, categorised as industry.

11

When was Switch released?

Switch was published in January 2021. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

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.