GPT3-6.7B + muP

Closed weights Microsoft,OpenAI 6.7B 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
Microsoft,OpenAI
Organisation type
Industry,Industry
Country
United States of America
Published
7 March 2022
Authors
Greg Yang, Edward J. Hu, Igor Babuschkin, Szymon Sidor, Xiaodong Liu, David Farhi, Nick Ryder, Jakub Pachocki, Weizhu Chen, Jianfeng Gao

What it does

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

Domain
Language
Task
Language modeling/generation

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
6.7B
Training data
300,000,000,000 tokens
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
1.3 × 10²² FLOP

"we outperform published numbers of the 6.7B GPT-3 model, with tuning cost only 7% of total pretraining cost." GPT-3 6.7B reported training compute is 1.2e+22 FLOP 1.2e+22 FLOP * 1.07 = 1.284e+22 FLOP 6 FLOP / parameter / token * 6660000000 parameters * 300000000000 tokens = 1.1988 × 10^22 FLOP

How it was established
Comparison with other models,Operation counting
Fine-tuning compute
8.4 × 10²⁰ FLOP

"we outperform published numbers of the 6.7B GPT-3 model, with tuning cost only 7% of total pretraining cost." GPT-3 6.7B reported training compute is 1.2e+22 FLOP 1.2e+22 FLOP * 0.07 = 8.4e+20 FLOP

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

their repo is open: https://github.com/microsoft/mup The technique is open, not the model. GPT-3 isn't open so it wouldn't be possible for people to recreate GPT-3 + muP with this code

How it is classified

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

Record confidence
Confident
Citations
262
Benchmark data
GPT3-6.7B + muP

Sources

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

Reference
Tensor Programs V: Tuning Large Neural Networks via Zero-Shot Hyperparameter Transfer
Last updated
25 May 2026

What the numbers mean

Background

GPT3-6.7B + muP was published by Microsoft,OpenAI, in the country recorded as United States of America, during March 2022. The publishing organisation is categorised as industry,Industry.

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

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

How it was trained

Producing it required arithmetic totalling around 1.3 × 10²² FLOP. That figure measures what producing the model cost, and has no bearing on how fast it answers.

Training consumed a corpus of around 300,000,000,000 tokens of text.

Answers

GPT3-6.7B + muP — common questions

01

GPT3-6.7B + muP— what is it used for?

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

02

GPT3-6.7B + muP— how much compute was used to train it?

Training consumed around 1.3 × 10²² FLOP. 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

GPT3-6.7B + muP— 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.

04

GPT3-6.7B + muP— is it open source?

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

05

GPT3-6.7B + muP— how many parameters does it have?

It has a parameter count of 6.7B. 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

GPT3-6.7B + muP— who created it?

It was published by Microsoft,OpenAI, based in United States of America, an organisation categorised as industry,Industry.

07

GPT3-6.7B + muP— when was it released?

It 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.

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.