Meena

Closed weights Google Brain 2.6B parameters January 2020

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 Brain
Organisation type
Industry
Country
United States of America
Published
28 January 2020
Authors
Dongling Xiao, Han Zhang, Yukun Li, Yu Sun, Hao Tian, Hua Wu, Haifeng Wang

What it does

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

Domain
Language
Task
Text autocompletion, Chat
Approach
Self-supervised learning

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
2.6B

"We present Meena, a multi-turn open-domain chatbot trained end-to-end on data mined and filtered from public domain social media conversations. This 2.6B parameter neural network is simply trained to minimize perplexity of the next token."

Training data
53,333,333,333 tokens

"The final Meena dataset contains 341GB of text (40B words)" Converting from GB to words yields 6.8e10, which is in the same OOM

Epochs
164
Batch size
82,655

61B tokens over 738k training steps, or 82655 tokens per batch on average. Not certain about warmup, etc

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
1.1 × 10²³ FLOP

https://arxiv.org/ftp/arxiv/papers/2104/2104.10350.pdf Table 4 In the paper: "We trained our best model for 30 days on a TPUv3 Pod (2,048 TPU cores) on the Meena dataset containing 40B words (or 61B BPE tokens) [...] by the end of training, the model had traversed the full training set 164 times (or epochs) and observed a total of about 10T tokens" Hardware: 30 * 24 * 3600 * (2048/2) * 1.23e14 * 0.3 = 9.794e22 Ops counting: 6 * 10T * 2.6B = 1.56E23 Geometric mean: sqrt(9.79e22*1.56E23) = 1.24e…

How it was established
Hardware,Operation counting,Third-party estimation

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
Chips used
1,024
Chip-hours
737,280
Wall-clock time
720 hours (30 days)

We trained our best model for 30 days on a TPUv3 Pod (2,048 TPU cores)

Hardware utilisation
HFU 34.3%

Per https://arxiv.org/ftp/arxiv/papers/2104/2104.10350.pdf 1.12e23 FLOPs used 1024 TPUv3s for 30 days: 30 * 24 * 3600 * 1024 * 1.23e14 = 3.2647e23 1.12e23 / 3.2647e23 HFU = 0.3431

Power draw
942.7 kW
Compute cost
$214,810

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
SOTA improvement

"We also propose a human evaluation metric called Sensibleness and Specificity Average (SSA)... the full version of Meena (with a filtering mechanism and tuned decoding) scores 79% SSA, 23% higher in absolute SSA than the existing chatbots we evaluated"

Record confidence
Confident
Citations
1,020

Sources

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

Reference
Towards a Human-like Open-Domain Chatbot
Last updated
25 May 2026

What the numbers mean

About this model

Meena was published by Google Brain, in United States of America, in January 2020. It comes out of industry.

It works in Language, and is recorded as doing text autocompletion, Chat.

This is a closed model: the trained values stayed with whoever produced them, and there is no local version to run.

Training and provenance

The training run consumed about 1.1 × 10²³ FLOP, on Google TPU v3. That figure describes the cost of creating it and has no bearing on how quickly it generates text.

Around 53,333,333,333 tokens went into training it.

It is tracked in the underlying dataset for one reason in particular: sOTA improvement.

Answers

Meena — common questions

01

Who created Meena?

Meena was published by Google Brain, based in United States of America, categorised as industry.

02

When was Meena released?

Meena was published in January 2020. 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

What is Meena used for?

Meena works in Language, and is recorded as handling text autocompletion, Chat. A model can carry several of each, so these are the areas it was built for rather than a limit on what it will attempt.

04

How much compute was used to train Meena?

Around 1.1 × 10²³ FLOP, on 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

What GPU do I need to run Meena?

None. Meena 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

Is Meena open source?

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

07

How many parameters does Meena have?

Meena has 2.6B parameters. "We present Meena, a multi-turn open-domain chatbot trained end-to-end on data mined and filtered from public domain social media conversations. This 2.6B parameter neural network is simply trained to minimize perplexity of the next token.". 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.