FLAN 137B

Closed weights Google Research 137B parameters September 2021

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 Research
Organisation type
Industry
Country
United States of America
Published
3 September 2021
Authors
Jason Wei, Maarten Bosma, Vincent Y. Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M. Dai, and Quoc V. Le

What it does

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

Domain
Language
Task
Language modeling, Question answering, Language modeling/generation
Approach
Self-supervised learning
Base model
LaMDA

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
137B

Abstract: "We take a 137B parameter pretrained language model and instruction tune it on over 60 NLP datasets verbalized via natural language instruction templates. We evaluate this instruction-tuned model, which we call FLAN, on unseen task types." Many models seem to be using the same 137B base transformer model?

Training data
2,490,000,000,000 tokens

"Model architecture and pretraining. In our experiments, we use LaMDA-PT, a dense left-to-right, decoder-only transformer language model of 137B parameters (Thoppilan et al., 2022). This model is pretrained on a collection of web documents (including those with computer code), dialog data, and Wikipedia, tokenized into 2.49T BPE tokens with a 32k vocabulary using the SentencePiece library (Kudo & Richardson, 2018). Around 10% of the pretraining data was non-English. Note that LaMDA-PT only has …

Epochs
1

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 × 10²⁴ FLOP

From section 2.4: Pretraining was done over 2.49T tokens. 6 * 2.49T * 137B = 2.047e24 Also, "instruction tuning takes around 60 hours on a TPUv3 with 128 cores." 128 TPUv3 cores = 64 TPUv3 chips. Environmental considerations section claims this took less than 2% of total time 1.23e14 * 64 * 60 * 3600 * 0.3 = 5.10e20

How it was established
Operation counting

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
Chip-hours
3,840

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.

Frontier model
Yes
Likely above 10²³ FLOP
Yes
Why it is tracked
Highly cited,SOTA improvement

Abstract: "FLAN substantially improves the performance of its unmodified counterpart and surpasses zero-shot 175B GPT-3 on 20 of 25 datasets that we evaluate." "FLAN even outperforms few-shot GPT-3 by a large margin on ANLI, RTE, BoolQ, AI2-ARC, OpenbookQA, and StoryCloze" SOTA is reported among unsupervised models

Record confidence
Confident
Citations
5,011

Sources

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

Reference
Finetuned Language Models Are Zero-Shot Learners
Last updated
25 May 2026

What the numbers mean

Where it came from

FLAN 137B was published by Google Research, in the country recorded as United States of America, during September 2021. It comes out of an organisation categorised as industry.

It works in the domain of Language, and is recorded as performing the task of language modeling, Question answering, Language modeling/generation.

Rather than being trained from scratch, it is derived from LaMDA. Most models at this scale are adapted from an existing base rather than built from nothing.

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

How it was trained

The training run consumed about 2 × 10²⁴ FLOP, on hardware recorded as Google TPU v3. That figure measures what producing the model cost, and has no bearing on how fast it answers.

Training consumed a corpus of around 2,490,000,000,000 tokens of text.

The reason it appears in this catalogue at all: highly cited,SOTA improvement.

Answers

FLAN 137B — common questions

01

FLAN 137B— who created it?

It was published by Google Research, based in United States of America, an organisation categorised as industry.

02

FLAN 137B— when was it released?

It was published in September 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.

03

FLAN 137B— what is it used for?

It works in the domain of Language, and is recorded as handling the task of language modeling, Question answering, Language modeling/generation. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

04

FLAN 137B— how much compute was used to train it?

Training consumed around 2 × 10²⁴ FLOP, on hardware recorded as 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.

05

FLAN 137B— what GPU do I need to run it?

None. This 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.

06

FLAN 137B— is it open source?

No. Its weights have not been published, so it exists only as a service controlled by its owner.

07

FLAN 137B— how many parameters does it have?

It has a parameter count of 137B. Abstract: "We take a 137B parameter pretrained language model and instruction tune it on over 60 NLP datasets verbalized via natural language instruction templates. We evaluate this instruction-tuned model, which we call FLAN, on unseen task types." Many models seem to be using the same 137B base transformer model?. 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.

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.