Google · Tools

Google Imagen

Visit website

At a glance

Price
Pricing on request
Vendor
Google

Specifications & properties

Key decision factors

Pricing model
Paid 1
Commercial usage rights
Enterprise with indemnification 1
Copyright status
User owns rights 1
GDPR / EU hosting
EU option available 1

Pricing

Free tier
No 1

Integration

API available
Yes 1

Quality

Maximum resolution
2K 1

Capabilities

Style range
Photorealism, impressionism, abstract and illustration styles 1
Text in image
Yes 1
Inpainting / editing
Yes 1

Compliance

C2PA / watermark
Invisible watermark 1
Report data / suggest a correction

Imagen is Google's family of high-fidelity text-to-image models, offered to developers through the Gemini API and Vertex AI and surfaced in Google consumer products such as Gemini and Whisk. Imagen 4 generates images up to 2K resolution and embeds Google's SynthID watermark in every generated image.

Profile

Imagen is Google's family of high-fidelity text-to-image models, accessed by developers through the Gemini API and Vertex AI and surfaced to consumers inside products such as Gemini and Whisk. Its defining trait is provenance: Imagen 4 embeds Google's SynthID watermark in every image it generates, not as an optional setting.

Who builds it

Imagen is built by Google, and the model family sits at the center of two different distribution paths at once. Developers reach it through the Gemini API or through Vertex AI on Google Cloud, while everyday users encounter the same underlying model line inside consumer-facing Google products like Gemini and Whisk, without necessarily knowing they are using Imagen directly. That split — an infrastructure-level API product that also quietly powers mainstream consumer features — is unusual among image generators, most of which ship as a single destination product.

What it does

Imagen 4 generates images up to 2K resolution, according to Google's own Gemini API documentation and DeepMind's model page, and Google documents a style range spanning photorealism, impressionism, abstract work and illustration. Text rendered inside a generated image is supported, and Google Cloud's documentation describes AI-based image editing on Vertex AI, so Imagen is not limited to one-shot generation from a prompt. For teams with data-residency requirements, Google's own documentation on Vertex AI data residency confirms an EU hosting option.

Pricing

Imagen is a paid product with no free tier: Google's Gemini API pricing page confirms usage is billed rather than offered free, and there is no published free allowance in Google's own materials. Access runs through the same Gemini API and Vertex AI billing that covers Google's other generative AI products, rather than a separate Imagen-specific plan. That puts Imagen squarely in developer/enterprise territory: there is no consumer-style free trial tier of the kind offered by freemium image generators, even though individual consumer products built on top of Imagen, such as Whisk, may apply their own separate access terms.

Rights and provenance

Google Cloud's Service Specific Terms and its dedicated generative-AI indemnification terms position commercial use of Imagen output as an indemnified enterprise offering, and Google's own terms confirm that ownership of generated output sits with the user rather than with Google. Provenance is where Imagen stands apart from most competitors: every image Imagen 4 generates carries an invisible SynthID watermark, according to Google's own Gemini API documentation and DeepMind's SynthID page, applied automatically rather than as something a user has to enable. Combined with indemnified commercial terms, that makes Imagen a model built for enterprise and platform-scale use, where being able to prove an image is AI-generated — and having Google stand behind commercial use of the output — both matter as much as raw image quality.

Frequently asked questions

What does Google Imagen cost?

Google Imagen is a paid product billed through the Gemini API, with no free tier, according to Google's own Gemini API pricing page. There is no separate Imagen-only plan — usage is billed as part of Gemini API or Vertex AI consumption.

Can I use Google Imagen images commercially, and who owns them?

Yes. Google Cloud's Service Specific Terms and its generative-AI indemnification terms position commercial use of Imagen output as an indemnified enterprise offering, and Google's own terms confirm the user owns the generated output rather than Google.

Is Google Imagen free?

No. Google's Gemini API pricing page confirms there is no free tier for Imagen; access is billed usage through the Gemini API or Vertex AI.

What resolution and styles can Google Imagen produce?

Imagen 4 generates images up to 2K resolution, according to Google's Gemini API documentation and DeepMind's Imagen model page, with a documented style range spanning photorealism, impressionism, abstract work and illustration.

Google Imagen vs FLUX: which should I choose?

Choose Google Imagen if you want a fully managed model tied to Google's Gemini API and Vertex AI, with an automatic SynthID watermark on every image and an EU hosting option — but note it has no free tier. Choose FLUX from Black Forest Labs if you want an open-weight option you can self-host: FLUX is also a paid, pay-per-use product with no free tier for the hosted API, but it additionally ships downloadable models such as FLUX.1 [dev].

Does Google Imagen label its AI images?

Yes. Every image generated by Imagen 4 carries an invisible SynthID watermark automatically, according to Google's own Gemini API documentation and DeepMind's SynthID page, without the user needing to enable it.