AlphaMissense

Closed weights Google DeepMind 93M parameters September 2023

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
Google DeepMind
Organisation type
Industry
Country
United States of America
Published
22 September 2023
Authors
Jun Cheng, Guido Novati, Joshua Pan, Clare Bycroft, Akvile ̇Žemgulyte ̇, Taylor Applebaum, Alexander Pritzel, Lai Hong Wong, Michal Zielinski, Tobias Sargeant, Rosalia G. Schneider,Andrew W. Senior, John Jumper, Demis Hassabis, Pushmeet Kohli,Žiga Avsec

What it does

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

Domain
Biology
Task
Protein pathogenicity prediction, Protein folding prediction, Proteins
Base model
AlphaFold 2
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
93M

"The model architecture is similar to that of AlphaFold (21), with minor modifications" Reference is to the AlphaFold 2 paper; that model had 93 million parameters

Training data
2,304,000,000 tokens

7800000 samples - size of training dataset (see Table S4 in supplementary materials) +1,345,605 variants for fine-tuning (but less could be used) see Table S1 around 9000000 samples is quite confident estimation

Epochs
4

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
Open source

Apache for code. weights not released https://github.com/google-deepmind/alphamissense

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

"By combining structural context and evolutionary conservation, our model achieves state-of-the-art results across a wide range of genetic and experimental benchmarks, all without explicitly training on such data." [Abstract]

Record confidence
Likely
Citations
1,226

Sources

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

Reference
Accurate proteome-wide missense variant effect prediction with AlphaMissense
Last updated
1 January 2026

What the numbers mean

Where it came from

AlphaMissense was published by Google DeepMind, in United States of America, in September 2023. industry is the category the publisher falls under.

It works in Biology, and is recorded as doing protein pathogenicity prediction, Protein folding prediction, Proteins.

It builds on AlphaFold 2, which is why it shares that model's general shape and size.

Because the weights are not available, none of the hardware figures elsewhere on this site apply to it.

Training and provenance

Around 2,304,000,000 tokens went into training it.

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

Answers

AlphaMissense — common questions

01

When was AlphaMissense released?

AlphaMissense was published in September 2023. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

02

What is AlphaMissense used for?

AlphaMissense works in Biology, and is recorded as handling protein pathogenicity prediction, Protein folding prediction, Proteins. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

03

What GPU do I need to run AlphaMissense?

None. AlphaMissense 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 AlphaMissense open source?

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

05

How many parameters does AlphaMissense have?

AlphaMissense has 93M parameters. "The model architecture is similar to that of AlphaFold (21), with minor modifications" Reference is to the AlphaFold 2 paper; that model had 93 million 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.

06

Who created AlphaMissense?

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

Source

Original publication

Record last updated 1 January 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.