Imagen
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
- 23 May 2022
- Authors
- Chitwan Saharia, William Chan, Saurabh Saxena, Lala Li, Jay Whang, Emily Denton, Seyed Kamyar Seyed Ghasemipour, Burcu Karagol Ayan, S. Sara Mahdavi, Rapha Gontijo Lopes, Tim Salimans, Jonathan Ho, David J Fleet, Mohammad Norouzi
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Image generation
- Task
- Text-to-image, Image generation
- 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
- 7.8B
- Training data
- tokens
2B 64x64 generation model, 600M 64->256 super-resolution model, 400M 256->1024 super-resolution model Uses encodings from a frozen T5-XXL, which should be included in total parameter count. Loading the model directly, there are 4,762,310,656 parameters in the encoder. 2B + 4.762B + 600M + 400M = 7.762 billion here they claim it is 3B parameters: https://arxiv.org/pdf/2407.15811
[IMAGE-TEXT PAIRS] "We train on a combination of internal datasets, with ≈ 460M image-text pairs, and the publicly available Laion dataset [61], with ≈ 400M image-text pairs."
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
- 1.5 × 10²² FLOP
- How it was established
- Hardware
256 TPU v4 chips for 64x64, for 4 days 128 TPU v4 chips for 64->256, for 2 days 128 TPU v4 chips for 256->1024, for 2 days 256 TPUs * 275 teraFLOPS/TPU * 4 days + 2 * (128 TPUs * 275 teraFLOPS/TPU * 2 days) * 40% utilization = 1.46e+22 FLOP
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
- 256
- Chip-hours
- 24,576
- Wall-clock time
- 96 hours
- Power draw
- 174.7 kW
- Compute cost
- $7,916
4 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
- API access
- 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
- Why it is tracked
- Significant use,SOTA improvement,Highly cited
- Record confidence
- Likely
- Citations
- 8,259
"Imagen achieves a new state-of-the-art FID score of 7.27 on the COCO dataset, without ever training on COCO" Also SOTA on MS-COCO (Table 5: https://arxiv.org/pdf/2206.10789v1)
Sources
Where this record came from and when it was last checked.
- Reference
- Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding
- Last updated
- 25 May 2026
What the numbers mean
Where it came from
Imagen was published by Google Brain, in the country recorded as United States of America, during May 2022. It comes out of an organisation categorised as industry.
It works in the domain of Image generation, and is recorded as performing the task of text-to-image, Image generation.
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 a computation budget of roughly 1.5 × 10²² FLOP, on hardware recorded as Google TPU v4. That figure measures what producing the model cost, and has no bearing on how fast it answers.
It is tracked in the underlying dataset for one reason in particular: significant use,SOTA improvement,Highly cited.
Answers
Imagen — common questions
Imagen— what is it used for?
It works in the domain of Image generation, and is recorded as handling the task of text-to-image, Image generation. These are the areas it was designed around; they describe intent rather than a hard boundary.
Imagen— how much compute was used to train it?
Training consumed around 1.5 × 10²² FLOP, on hardware recorded as 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.
Imagen— 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.
Imagen— is it open source?
No. Its weights have not been published, so it exists only as a service controlled by its owner.
Imagen— how many parameters does it have?
It has a parameter count of 7.8B. 2B 64x64 generation model, 600M 64->256 super-resolution model, 400M 256->1024 super-resolution model Uses encodings from a frozen T5-XXL, which should be included in total parameter count. Loading the model directly, there are 4,762,310,656 parameters in the encoder. 2B + 4.762B + 600M + 400M = 7.762 billion here they claim it is 3B parameters: https://arxiv.org/pdf/2407.15811. 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.
Imagen— who created it?
It was published by Google Brain, based in United States of America, an organisation categorised as industry.
Imagen— when was it released?
It was published in May 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.
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.