Aeneas

Closed weights Google DeepMind,University of Nottingham,University of Warwick,Athens University of Economics and Business,Google,University of Oxford July 2025

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,University of Nottingham,University of Warwick,Athens University of Economics and Business,Google,University of Oxford
Organisation type
Industry,Academia,Academia,Academia,Industry,Academia
Country
United States of America, United Kingdom of Great Britain and Northern Ireland, Greece
Published
23 July 2025
Authors
Yannis Assael, Thea Sommerschield, Alison Cooley, Brendan Shillingford, John Pavlopoulos, Priyanka Suresh, Bailey Herms, Justin Grayston, Benjamin Maynard, Nicholas Dietrich, Robbe Wulgaert, Jonathan Prag, Alex Mullen, Shakir Mohamed

What it does

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

Domain
Vision, Multimodal, Language
Task
Character recognition (OCR), Visual question answering

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

"batch size of 1,024 text–image pairs, using the LAMB81 optimizer. The learning rate follows a schedule with a peak value of 3 × 10−3, a warm-up phase of 4,000 steps and a total of 1 million steps." -> ~10^9 image-text 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
2.3 × 10²¹ FLOP

197000000000000 FLOP / chip / sec * 64 chips * 168 hours * 3600 sec / hour * 0.3 [assumed utilization] = 2.2875955e+21 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 v5e
Chips used
64
Wall-clock time
168 hours (7 days)

"The model was trained for one week using 64 Tensor Processing Unit v5e chips on the Google Cloud platform"

Power draw
28.2 kW
Cloud vendor
Google Cloud

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
Hosted access (no API)
Training code
Open source

Apache 2.0 https://github.com/google-deepmind/predictingthepast/tree/main/train

How it is classified

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

Record confidence
Confident

Sources

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

Reference
Contextualizing ancient texts with generative neural networks
Last updated
28 November 2025

What the numbers mean

About this model

Aeneas was published by Google DeepMind,University of Nottingham,University of Warwick,Athens University of Economics and Business,Google,University of Oxford, in United States of America, in July 2025. industry,Academia,Academia,Academia,Industry,Academia is the category the publisher falls under.

It works in Vision, Multimodal, Language, and is recorded as doing character recognition (OCR), Visual question answering.

Its weights were never published, so it can only be reached through its provider. No graphics card changes that.

What went into building it

Producing it required around 2.3 × 10²¹ FLOP of arithmetic, on Google TPU v5e, which is a statement about the training budget rather than about inference.

Answers

Aeneas — common questions

01

What is Aeneas used for?

Aeneas works in Vision, Multimodal, Language, and is recorded as handling character recognition (OCR), Visual question answering. These are the areas it was designed around; they describe intent rather than a hard boundary.

02

How much compute was used to train Aeneas?

Around 2.3 × 10²¹ FLOP, on Google TPU v5e. 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 Aeneas?

None. Aeneas 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 Aeneas open source?

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

05

How many parameters does Aeneas have?

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

06

Who created Aeneas?

Aeneas was published by Google DeepMind,University of Nottingham,University of Warwick,Athens University of Economics and Business,Google,University of Oxford, based in United States of America, categorised as industry,Academia,Academia,Academia,Industry,Academia.

07

When was Aeneas released?

Aeneas was published in July 2025.

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.