Imagen 4 Pricing: $0.02 to $0.06 an Image, and What Replaced It (2026)

Imagen 4 pricing was $0.02 an image for Fast, $0.04 for Standard and $0.06 for Ultra, but Google has now shut all three down in the Gemini API and Vertex AI. This guide turns those list prices into cost per 1,000 images, gives the shutdown dates, and prices the models Google moved Imagen users to, as Google listed them on 25 October 2026. We also ran two of those replacements on Wireflow, the node-based AI canvas we build, and report the credits and seconds each image took. We sell another way to run image models, so check Google's pages before you buy.

The short version: the closest live option is Google's Nano Banana 2.1, which the Gemini API lists as the replacement for every Imagen 4 model. It costs $0.0336 for a 1K image, which is 68% more than Imagen 4 Fast, 16% less than Imagen 4 and 44% less than Imagen 4 Ultra. Run as a batch, it is $0.0168, below all three.

Imagen 4 prices per image, by tier

Google charged one flat price per image for each Imagen 4 model, with no split by output size. These are the rates on Google Cloud's Vertex AI pricing page, which now calls the service Agent Platform, on 25 October 2026. The page still lists them even though the models are retired.

Imagen 4 model	Model ID	Price per image	1,000 images	10,000 images	Output sizes
Imagen 4 Fast	imagen-4.0-fast-generate-001	$0.02	$20	$200	Up to 1K (1024 x 1024, 1408 x 768)
Imagen 4	imagen-4.0-generate-001	$0.04	$40	$400	Up to 2K (2048 x 2048, 2816 x 1536)
Imagen 4 Ultra	imagen-4.0-ultra-generate-001	$0.06	$60	$600	Up to 2K (2048 x 2048, 2816 x 1536)
Imagen 4 upscaling	Separate call	$0.06	$60	$600	2K, 3K or 4K

Each request could return up to 4 images, and each image was billed, so a 4-image request on Imagen 4 Ultra cost $0.24. Upscaling a generated image was a separate $0.06 call, which doubled the cost of an Ultra image you planned to print. For how other image APIs charge, see our roundup of image generation APIs.

Can you still use Imagen 4?

Not from Google. Google released the three Imagen 4 models in the Gemini API on 24 June 2025 and retired them about 14 months later. These are the dates from Google's own lifecycle pages on 25 October 2026:

• Gemini API: imagen-4.0-generate-001, imagen-4.0-ultra-generate-001 and imagen-4.0-fast-generate-001 shut down on 17 August 2026. The two preview IDs shut down on 17 February 2026.
• Vertex AI: all three endpoints were discontinued on 30 June 2026, and the standard and Ultra models sit in Google's retired-models table.
• Google's Imagen docs: the Gemini API page now opens with "Imagen models are shut down. Use Nano Banana for image generation."

That is why search results still show Imagen 4 prices: the Google Cloud pricing table has not been updated, and many guides copy it. Some resellers also keep Imagen 4 model pages up; Replicate's google/imagen-4 page was still live on 25 October. Make a test call before you build on one. If you need a live endpoint today, our text-to-image API page shows one way to call several current models.

What Google moved Imagen users to, and what it costs

The Gemini API names Nano Banana 2.1 (gemini-nano-banana-2.1) as the replacement for every Imagen 4 model. Vertex AI's retired-models table points to gemini-2.5-flash-image, the first Nano Banana, but Google's deprecation tables retire that model on 15 March 2027 in both APIs, so it is a short stop.

Google prices these models by tokens rather than per image. Nano Banana 2.1 charges $30 per million image output tokens, and a 1K image uses 1,120 tokens, which is where $0.0336 comes from. These are the per-image figures from the Gemini API pricing page on 25 October 2026.

Google model	Model ID	1K image	2K image	4K image	Batch, 1K
Nano Banana 2.1	gemini-nano-banana-2.1	$0.0336	$0.0504	$0.113	$0.0168
Nano Banana 2 Lite	gemini-3.1-flash-lite-image	$0.0336	Not listed	Not listed	$0.0168
Nano Banana 2	gemini-3.1-flash-image	$0.067	$0.101	$0.151	$0.034
Nano Banana Pro	gemini-3-pro-image	$0.134	$0.134	$0.24	$0.067

Here is how each old tier compares by price:

• Imagen 4 Fast users: Nano Banana Lite and Nano Banana 2.1 both cost $0.0336 at 1K, more than the old $0.02. Batch brings either to $0.0168, just under it.
• Imagen 4 users: Nano Banana 2.1 at 1K is $0.0336, below the old $0.04. At 2K it is $0.0504, 26% more.
• Imagen 4 Ultra users: Nano Banana 2.1 at 2K costs less than Ultra did. Nano Banana Pro is $0.134 at 1K or 2K, more than twice the old Ultra price.
• Older Gemini image models: Google calls Nano Banana 2.1 an update to Nano Banana 2, and at 1K it costs half as much.

Running Google's replacements on Wireflow: cost and time from our run

Imagen 4 is not in Wireflow's model catalog; we checked on 25 October 2026. To price the replacement path, we ran one product photo prompt, a matte black mug on an oak table, through Nano Banana Lite and Nano Banana 2.1 at 1K, square, in one workflow that day. Every paid plan on the pricing page worked out to about 1.35 cents a credit: Starter is $36 for 2,660 credits, Professional $59 for 4,370 and Team $299 for 22,140.

Model	Credits per image, our run	Time, our run	Cost on Starter or Professional	Same image on Google's API
Nano Banana Lite	16	3.3 seconds	$0.22	$0.0336
Nano Banana 2.1 at 1K	20	11.2 seconds	$0.27	$0.0336

Both nodes ran in parallel, so the pair came back in 13.9 seconds for 36 credits. Times come from one run and will change with load, and the pricing page notes that resolution changes the cost, so check the estimate on the node before a 2K or 4K run. Turning credits into a price per image is the simplest way to compare an AI image generator that bills in credits with an API that bills in dollars.

In our single test, Nano Banana Lite added a coaster and coffee that the prompt did not ask for, while Nano Banana 2.1 kept closer to the prompt, with a matte finish and the eucalyptus beside the mug. For more on this kind of shot, see our page on AI product photography.

To be plain about price: Google's own API is the cheaper way to make single images. A 1K Nano Banana 2.1 image costs $0.0336 there and about $0.27 in our run, roughly eight times more. Our route fits when the image is one step in a longer job, such as a prompt, two image models, an upscale and a video, run together from one API call for the whole workflow, with no token math.

How to budget the move off Imagen 4

1. Start from your old bill. Multiply your monthly images by $0.02, $0.04 or $0.06 to see what Imagen 4 cost you.
2. Pick the output size you really use. Nano Banana 2.1 charges by size: $33.60 per 1,000 images at 1K, $50.40 at 2K and $113 at 4K.
3. Move work that can wait to batch. Batch halves the price, to $16.80 per 1,000 images at 1K.
4. Change the call, not just the name. Google's migration notes say to call generate_content instead of generate_images and read the image from the returned content parts.
5. Re-test your prompts. A new model reads prompts differently, so check product colours, text and framing on a sample before you switch a whole catalogue.

For example, a shop making 5,000 product images a month on Imagen 4 paid $200 at the standard rate. On Nano Banana 2.1 the same volume costs $168 at 1K, $84 in batch or $252 at 2K. For the code side of the switch, see our guide to calling Nano Banana from code.

Prices like these move often, so we date every figure. For another image model priced the same way, per image by size, see our Krea 3 pricing breakdown.

Try it yourself: Open this side-by-side workflow. The nodes are pre-configured: one product photo prompt goes into Nano Banana Lite and Nano Banana 2.1 at 1K, so you can compare the images and the credits each one uses. You can open and change this workflow for free. A new account gets up to 60 starter credits and a free video, and each run uses credits. Plans start at $36 a month for 2,660 credits.

FAQ

How much does Imagen 4 cost per image?

Google listed $0.02 for Imagen 4 Fast, $0.04 for Imagen 4 and $0.06 for Imagen 4 Ultra, plus $0.06 to upscale an image to 2K, 3K or 4K. You can no longer buy at those rates, because the models are shut down.

Is Imagen 4 still available?

No. The Gemini API shut down all three Imagen 4 models on 17 August 2026, and Vertex AI discontinued them on 30 June 2026. Google's Imagen page in the Gemini docs now says to use Nano Banana instead.

What replaced Imagen 4?

In the Gemini API, Google lists Nano Banana 2.1 (gemini-nano-banana-2.1) as the replacement for every Imagen 4 model. Vertex AI's table points to gemini-2.5-flash-image, which Google plans to retire on 15 March 2027.

Is Nano Banana 2.1 cheaper than Imagen 4?

It depends on the tier you used. At 1K it costs $0.0336, less than Imagen 4 at $0.04 and Ultra at $0.06 but more than Fast at $0.02. In batch mode it is $0.0168, below all three.

Why do some sites still show Imagen 4 prices?

Google Cloud's pricing page still lists the Imagen 4 rates, and many guides copy that table. Check Google's model lifecycle pages for the shutdown dates before you budget.

Can I still run Imagen 4 through a reseller?

Some resellers still have Imagen 4 pages, such as Replicate's google/imagen-4 page on 25 October 2026, but Google has shut the models down in its own APIs. Make a test call before you build on one.

Does Google charge for the text prompt too?

Yes, as input tokens. Nano Banana 2.1 input costs $1.50 per million tokens, so a prompt of a few dozen words adds well under a cent per image.

Conclusion

Imagen 4 cost $0.02 to $0.06 an image, and that price list now describes models you can no longer call. For most teams the like-for-like move is Nano Banana 2.1 at $0.0336 for a 1K image on Google's API, or half that in batch. If you want Google's image models as steps in a larger job, with the credits for each run shown on the node, you can build that on Wireflow, starting from our batch image generation API page.