Chinchilla

Closed weights DeepMind 70B parameters March 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
29 March 2022
Authors
Jordan Hoffmann, Sebastian Borgeaud, Arthur Mensch, Elena Buchatskaya, Trevor Cai, Eliza Rutherford, Diego de Las Casas, Lisa Anne Hendricks, Johannes Welbl, Aidan Clark, Tom Hennigan, Eric Noland, Katie Millican, George van den Driessche, Bogdan Damoc, Aurelia Guy, Simon Osindero, Karen Simonyan, Erich Elsen, Jack W. Rae, Oriol Vinyals, Laurent Sifre

What it does

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

Domain
Language
Task
Language modeling
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
70B

"We test this hypothesis by training a predicted compute-optimal model, \chinchilla, that uses the same compute budget as \gopher but with 70B parameters and 4× more more data. \chinchilla uniformly and significantly outperforms \Gopher (280B), GPT-3 (175B), Jurassic-1 (178B), and Megatron-Turing NLG (530B) on a large range of downstream evaluation tasks."

Training data
1,400,000,000,000 tokens

Table 1 shows Chinchilla was training on 1.4 trillion tokens 1 token ~ 0.75 words

Epochs
1
Batch size
3,000,000

Table 1. "1.5M → 3M"

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
5.8 × 10²³ FLOP

"Both Chinchilla and Gopher have been trained for the same number of FLOPs but differ in the size of the model and the number of training tokens." We see the number of flops in table 3

How it was established
Reported

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 v4,Google TPU v3

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

How it is classified

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

Foundation model
Yes
Likely above 10²³ FLOP
Yes
Why it is tracked
SOTA improvement,Historical significance

Proposes new scaling law, with good empirical results

Record confidence
Confident
Citations
3,188
Benchmark data
Chinchilla

Sources

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

Reference
Training Compute-Optimal Large Language Models
Last updated
25 May 2026

What the numbers mean

About this model

Chinchilla was published by DeepMind, in United Kingdom of Great Britain and Northern Ireland, in March 2022. The organisation is categorised as industry.

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

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 5.8 × 10²³ FLOP of arithmetic, on Google TPU v4,Google TPU v3, which is a statement about the training budget rather than about inference.

The training set ran to roughly 1,400,000,000,000 tokens.

Its inclusion criterion is sOTA improvement,Historical significance.

Answers

Chinchilla — common questions

01

How much compute was used to train Chinchilla?

Around 5.8 × 10²³ FLOP, on Google TPU v4,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.

02

What GPU do I need to run Chinchilla?

None. Chinchilla 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.

03

Is Chinchilla open source?

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

04

How many parameters does Chinchilla have?

Chinchilla has 70B parameters. "We test this hypothesis by training a predicted compute-optimal model, \chinchilla, that uses the same compute budget as \gopher but with 70B parameters and 4× more more data. \chinchilla uniformly and significantly outperforms \Gopher (280B), GPT-3 (175B), Jurassic-1 (178B), and Megatron-Turing NLG (530B) on a large range of downstream evaluation tasks.". 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.

05

Who created Chinchilla?

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

06

When was Chinchilla released?

Chinchilla was published in March 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.

07

What is Chinchilla used for?

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

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.