Muse is a text-to-image model that also supports editing images through its generative transformer approach.
Imagen
Google Research introduced Imagen as a text-to-image diffusion model in this research paper.
About Imagen
Imagen is a Google Research project page for a text-to-image diffusion model, presented in a research paper from the Google Brain Team. The model generates images from text descriptions. It pairs a large pretrained language model, T5-XXL, which is kept frozen, with a conditional diffusion model. The diffusion model first produces a 64×64 image, and separate super-resolution diffusion models upsample it to 256×256 and then 1024×1024.
The page reports that the paper's authors found that scaling the text encoder improved image fidelity and image-text alignment more than scaling the image diffusion model. It reports a COCO FID score of 7.27 for a model not trained on COCO. The authors also introduced DrawBench, a benchmark for side-by-side human evaluation of text-to-image models. In these comparisons, human raters preferred Imagen over VQ-GAN+CLIP, Latent Diffusion Models, and DALL-E 2. The page also links to related Imagen Video and Imagen Editor work. Pricing is not known, and the page does not describe a public interface.
Key features
- Generates images from text descriptions using a cascaded diffusion model
- Uses a frozen T5-XXL language model to encode input text
- Upsamples images from 64×64 to 256×256 and 1024×1024 resolution
- Introduces DrawBench, a benchmark for side-by-side text-to-image evaluation
- Includes a thresholding diffusion sampler for large classifier-free guidance weights
- Describes an Efficient U-Net architecture for lower compute and memory use
Best for: Researchers and readers who want to study text-to-image diffusion models and how they are evaluated.
Good for
- Reports a COCO FID score of 7.27 without training on COCO
- Human raters preferred its outputs over several compared models in side-by-side tests
- Provides DrawBench, a benchmark covering compositionality, spatial relations, and rare words
Watch out for
- The page presents research results and does not describe a public product interface
Imagen at a glance
- Website
- imagen.research.google
- Category
- AI & Machine Learning › Models & ML Research
- Type
- Website
- Pricing
- Not confirmed. Check the site
- Platforms
- Web
- Open source
- Not stated
- Language
- English
- Last checked
- Oct 6, 2026 · Online when checked
- Listed
- Oct 5, 2026
Imagen alternatives
Tools and sites you could use instead, ranked by how closely they match.
The ERNIE blog announces model releases and shares information about training and evaluation results.
Stable Diffusion v1.4 is a text-to-image model documented with its model card on Hugging Face.
TokenVerse presents research on personalizing image generation through multi-concept token modulation.
This research paper presents Word-As-Image, a method for incorporating semantic meaning into typography.
Image Arena ranks text-to-image generators and lets users compare sample outputs side by side.
Popular in Models & ML Research
Well-known listings from the same topic.
Falcon LLMfalconllm.tii.aewTAObittensor.comAlpacacrfm.stanford.edu/2023/03/13/alpaca.htmlSmolVLAhuggingface.co/blog/smolvlaAI Indexhai.stanford.edu/ai-index
Imagen FAQ
How did Imagen compare with other models in the paper?
In side-by-side human comparisons, raters preferred Imagen over VQ-GAN+CLIP, Latent Diffusion Models, and DALL-E 2, for both sample quality and image-text alignment.
Is something wrong or out of date?
Tell us and an editor will check it.
Own this website?
Claim this listing
Claim the listing and suggest updates yourself. An editor checks every claim first.