AlphaTensor

Closed weights DeepMind October 2022

No estimate

No hardware requirements for this model

The weights for this model have not been published, so it cannot be downloaded or run on your own hardware at any size. It is reachable only through its provider, and no graphics card changes that.

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
DeepMind
Organisation type
Industry
Country
United Kingdom of Great Britain and Northern Ireland
Published
5 October 2022
Authors
Alhussein Fawzi, Matej Balog, Aja Huang, Thomas Hubert, Bernardino Romera-Paredes, Mohammadamin Barekatain, Alexander Novikov, Francisco J. R. Ruiz, Julian Schrittwieser, Grzegorz Swirszcz, David Silver, Demis Hassabis & Pushmeet Kohli

What it does

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

Domain
Other, Games, Mathematics
Approach
Supervised

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.

Training data
tokens

"We create a dataset containing 5 million such tensor-factorization pairs."

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
7.1 × 10²⁰ FLOP

Compute: 0.3 [assumed utilization] * (64 cores / 2 cores per chip )*123000000000000 FLOP / TPU v3 chip / sec * 7 days * 24 hours / day * 3600 sec / hour = 7.1414784e+20 FLOP

How it was established
Hardware

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
64
Wall-clock time
168 hours (7 days)

In practice, the procedure takes a week to converge.

Power draw
57.6 kW

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
Closed — provider access only
Model access
Unreleased
Training code
Unreleased

Apache 2.0 for algorithmic outputs https://github.com/google-deepmind/alphatensor

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

Sources

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

Reference
Discovering faster matrix multiplication algorithms with reinforcement learning
Last updated
28 November 2025

What the numbers mean

What this model is

AlphaTensor was published by DeepMind, in United Kingdom of Great Britain and Northern Ireland, in October 2022. industry is the category the publisher falls under.

It works in Other, Games, Mathematics.

This is a closed model: the trained values stayed with whoever produced them, and there is no local version to run.

How it was trained

Training it took roughly 7.1 × 10²⁰ FLOP of computation, on Google TPU v3 — a measure of what producing the model cost, not of how fast it answers.

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

Answers

AlphaTensor — common questions

01

What is AlphaTensor used for?

AlphaTensor works in Other, Games, Mathematics. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

02

How much compute was used to train AlphaTensor?

Around 7.1 × 10²⁰ 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.

03

What GPU do I need to run AlphaTensor?

None. AlphaTensor is a closed model — its weights were never published, so it cannot be downloaded or run on your own hardware at any price. It is reachable only through its provider.

04

Is AlphaTensor open source?

No. AlphaTensor has not had its weights published, so it exists only as a service controlled by its owner.

05

How many parameters does AlphaTensor have?

No parameter count has been published for AlphaTensor, which is why no memory or speed figure appears on this page.

06

Who created AlphaTensor?

AlphaTensor was published by DeepMind, based in United Kingdom of Great Britain and Northern Ireland, categorised as industry.

07

When was AlphaTensor released?

AlphaTensor was published in October 2022. 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 28 November 2025

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.