Google Discontinues Imagen 4: Developers Compelled to Transition to Gemini
Google has officially discontinued its three primary Imagen 4 image generation model endpoints as of August 17, 2026. This move compels developers to transition their apps and pipelines to the more recent Gemini 3.1 Flash Image model or risk severe service disruptions.
The three deprecated model IDs — imagen-4.0-generate-001, imagen-4.0-ultra-generate-001, and imagen-4.0-fast-generate-001 — are no longer functional from today onwards. Google initially announced the deprecation in June 2026, providing developers with approximately two months to prepare. However, the transition is not a straightforward replacement: the generate_images() API method has been entirely removed, and image generation must now be executed via the generate_content() method. This change necessitates developers to update their codebases and re-test their workflows.
Developers who are still utilizing Imagen 4 endpoints are recommended to promptly transition to gemini-3.1-flash-image. It is crucial to thoroughly test prompt adherence, aspect ratios, SynthID watermarking, latency, and API quotas before resuming production traffic. Google’s Firebase AI Logic documentation also verifies that all support for Imagen models has now ceased, with the Gemini “Nano Banana” image model family being the official replacement.
Source: Google AI for Developers – Gemini API Release Notes & Deprecation Changelog