Google is shutting down the Imagen 4 Standard, Ultra and Fast image-generation endpoints in the Gemini API on August 17, 2026, ending access to those model endpoints for developers using them through Google’s Gemini developer platform.
The company announced the deprecation on June 15 and advised developers to migrate their integrations to newer Gemini image-generation models to avoid service interruptions. Google’s current deprecation records identify imagen-4.0-generate-001, imagen-4.0-ultra-generate-001 and imagen-4.0-fast-generate-001 as the affected endpoints.
The change does not represent an end to Google’s image-generation technology. Instead, Google is retiring the Imagen 4 endpoints available through the Gemini API and moving developers toward its Gemini-native image models, which the company now groups under the Nano Banana family.
According to Google’s Gemini API deprecation schedule, a model marked for shutdown is eventually completely turned off and its endpoint is no longer available. The same documentation lists gemini-3.1-flash-image as the recommended replacement for the three Imagen 4 endpoints.
Google’s Imagen 4 model documentation also states that the Standard, Ultra and Fast endpoints are deprecated and will be shut down on August 17, 2026. The documentation tells developers to migrate to Gemini 3.1 Flash Image to avoid service interruptions.
The three Imagen 4 models were introduced as a family designed to give developers different balances of image quality, speed and cost. Google announced the general availability of Imagen 4 and Imagen 4 Ultra in the Gemini API and Google AI Studio on August 15, 2025, alongside Imagen 4 Fast, which was introduced for faster image generation and high-volume workloads.
At launch, Google priced Imagen 4 Fast at $0.02 per output image. The company later listed Imagen 4 at $0.04 per image and Imagen 4 Ultra at $0.06 per image in its Gemini API pricing documentation.
Google’s current image-generation documentation shows a different model lineup. Gemini’s image-generation guide identifies Gemini 3.1 Flash Image, also known as Nano Banana 2, as the general-purpose image model recommended for balancing performance, cost and latency. Gemini 3.1 Flash Lite Image, or Nano Banana 2 Lite, is positioned as the more efficient option, while Gemini 3 Pro Image, known as Nano Banana Pro, is aimed at professional asset production and complex instructions.
Gemini 3.1 Flash Image supports image generation and editing, with output options ranging from 512 pixels to 4K. Google’s documentation also lists support for multiple reference images, improved text rendering and, on the current model, grounding with Google Search and Google Image Search. The model’s documentation says it is designed for speed and high-volume developer workloads.
Google’s current documentation also says Nano Banana is the name for Gemini’s native image-generation capabilities. The company describes Gemini as able to generate, edit and refine images conversationally using text, images or a combination of both.
The migration also requires more than changing a model identifier in some integrations. Google’s Imagen migration documentation describes changes to the API workflow, including moving from the dedicated image-generation method to Gemini’s content-generation approach and handling image data through returned content parts rather than a dedicated image response object.
For developers moving from Imagen, that means production applications need to account for differences in model names, request methods and response handling. Google recommends testing the replacement model before the Imagen endpoint is shut down.
Pricing is another consideration. Google’s current Gemini API pricing lists Gemini 3.1 Flash Image at approximately $0.045 for a 512-pixel image, $0.067 for 1K, $0.101 for 2K and $0.151 for 4K output under standard pricing. Gemini 3.1 Flash Lite Image is listed at approximately $0.0336 for 1K output under standard pricing.
Those figures mean developers should review their image-generation costs rather than assuming that a migration will preserve the pricing structure of Imagen 4. The appropriate replacement can also differ by workload, even though Google specifically recommends Gemini 3.1 Flash Image for the Imagen 4 shutdown.
Google has been making similar changes across its image-model lineup. Its Gemini API records show that preview versions of Gemini 3.1 Flash Image and Gemini 3 Pro Image were shut down on June 25, 2026, with their stable counterparts becoming the replacements. Earlier, Google retired the Gemini 2.5 Flash Image preview endpoint and moved developers to its stable image model.
The sequence shows Google consolidating its developer-facing image-generation services around stable Gemini models rather than maintaining separate generations of Imagen and Gemini image endpoints indefinitely.
For developers affected by today’s shutdown, the immediate task is to identify applications still calling the three retired Imagen 4 model IDs and move those workloads to Google’s current Gemini image-generation models. Google’s official documentation identifies gemini-3.1-flash-image as the direct recommended replacement.
Discover more from Aree Blog
Subscribe now to keep reading and get access to the full archive.
