CoCa
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
- 14 June 2022
- Authors
- Jiahui Yu, Zirui Wang, Vijay Vasudevan, Legg Yeung, Mojtaba Seyedhosseini, Yonghui Wu
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Vision
- Task
- Image classification, Visual question answering, Image captioning
- 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
- 2.1B
- Training data
- 1,351,680,000,000 tokens
- Epochs
- 6.88
"Our largest CoCa model ("CoCa" in short) follows the ViT-giant setup in [21] with 1B-parameters in the image encoder and 2.1B-parameters altogether with the text decoder"
JFT is 3 billion captioned images, ALIGN is 1.8 billion captioned images "we use a batch size of 65,536 image-text pairs, where half of each batch comes from JFT and ALIGN, respectively. All models are trained on the combined contrastive and captioning objectives in Eq.(4) for 500k steps, roughly corresponding to 5 epochs on JFT and 10 epochs on ALIGN." (5*3b+10*1.8b)/4.8b=6.875 epochs on average
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
- 7.3 × 10²² FLOP
- How it was established
- Hardware
"Pretraining CoCa takes about 5 days on 2,048 CloudTPUv4 chips" 275 teraFLOP/s * 2048 * 5 * 24 * 3600 * 0.3 (assumed utilization) = 7.3e22
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
- Chips used
- 2,048
- Wall-clock time
- 120 hours
- Power draw
- 1.4 MW
- Compute cost
- $78,043
5 days
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.
- Why it is tracked
- SOTA improvement
- Record confidence
- Confident
- Citations
- 1,719
"Notably on ImageNet classification, CoCa obtains 86.3% zero-shot top-1 accuracy, 90.6% with a frozen encoder and learned classification head, and new state-of-the-art 91.0% top-1 accuracy on ImageNet with a finetuned encoder."
Sources
Where this record came from and when it was last checked.
- Reference
- CoCa: Contrastive Captioners are Image-Text Foundation Models
- Last updated
- 25 May 2026
What the numbers mean
Background
CoCa was published by Google Research, in United States of America, in June 2022. It comes out of industry.
It works in Vision, and is recorded as doing image classification, Visual question answering, Image captioning.
This is a closed model: the trained values stayed with whoever produced them, and there is no local version to run.
Training and provenance
Training it took roughly 7.3 × 10²² FLOP of computation, on Google TPU v4 — a measure of what producing the model cost, not of how fast it answers.
Around 1,351,680,000,000 tokens went into training it.
Its inclusion criterion is sOTA improvement.
Answers
CoCa — common questions
How much compute was used to train CoCa?
Around 7.3 × 10²² FLOP, on Google TPU v4. 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.
What GPU do I need to run CoCa?
None. CoCa 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.
Is CoCa open source?
No. CoCa has not had its weights published, so it exists only as a service controlled by its owner.
How many parameters does CoCa have?
CoCa has 2.1B parameters. "Our largest CoCa model ("CoCa" in short) follows the ViT-giant setup in [21] with 1B-parameters in the image encoder and 2.1B-parameters altogether with the text decoder". 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.
Who created CoCa?
CoCa was published by Google Research, based in United States of America, categorised as industry.
When was CoCa released?
CoCa was published in June 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.
What is CoCa used for?
CoCa works in Vision, and is recorded as handling image classification, Visual question answering, Image captioning. 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.
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.