Gemini image-generation model shutdown: what to do if your app is broken on 17 August 2026
Three Imagen 4 API endpoints shut down permanently on 17 August 2026. Here's what broke, why, and how to migrate to gemini-3.1-flash-image.
On 17 August 2026, Google permanently shut down three Imagen 4 image-generation model IDs on the Gemini API. If your application calls any of them, it is now returning hard errors. This is not a deprecation warning. The endpoints are gone.
The three retired model IDs are:
imagen-4.0-generate-001imagen-4.0-ultra-generate-001imagen-4.0-fast-generate-001
All three were released on 24 June 2025 and deprecated on 15 June 2026, giving developers a two-month migration window. That window closed on 17 August 2026.
What broke and why
Google has been consolidating the Gemini model family for over a year. All Gemini 1.0 and 1.5 models are already retired. On 1 June 2026, the entire Gemini 2.0 Flash cluster, including gemini-2.0-flash, gemini-2.0-flash-001, gemini-2.0-flash-lite, and gemini-2.0-flash-lite-001, was shut down. The August 2026 image model shutdown continues that pattern, moving developers from the standalone Imagen 4 model IDs onto the native Gemini image generation models.
The preview variants followed a slightly different schedule. gemini-3.1-flash-image-preview and gemini-3-pro-image-preview were shut down on 25 June 2026, ahead of the stable model shutdown.
Where to migrate
Google’s official deprecations page lists a single recommended replacement for all three retired Imagen 4 IDs: gemini-3.1-flash-image.
This model reached general availability on 28 May 2026 and has no announced shutdown date. A higher-capability counterpart, gemini-3-pro-image, is also generally available from the same date and maps to the former gemini-3-pro-image-preview.
A quick reference for the full migration path:
| Retired model ID | Replacement |
|---|---|
imagen-4.0-generate-001 | gemini-3.1-flash-image |
imagen-4.0-ultra-generate-001 | gemini-3.1-flash-image |
imagen-4.0-fast-generate-001 | gemini-3.1-flash-image |
gemini-3.1-flash-image-preview | gemini-3.1-flash-image |
gemini-3-pro-image-preview | gemini-3-pro-image |
One important caveat: this is not a drop-in string replacement. The GA image models use a different API surface from the Imagen 4 endpoints, so you will need to review your generate_images() calls and adjust parameters accordingly. Check the Gemini API models guide for the specifics of each model’s request format.
What this means for your production code
If you are seeing errors right now, the first step is to grep your codebase for any of the three retired model ID strings and replace them. Then verify that your integration code matches the request format expected by gemini-3.1-flash-image.
A few things worth knowing as you migrate:
Avoid unversioned aliases. Google has, in some cases, repointed a deprecated preview ID to a newer model after shutdown rather than returning an error. This is unreliable behaviour. Always use explicit, versioned model IDs in production code so you know exactly what you are calling and when the next migration deadline arrives.
Mobile and Firebase developers should use Remote Config. If your app calls image generation models directly from a mobile client, Google recommends using Firebase Remote Config or server-side prompt templates to manage model IDs. That way you can update the model name without shipping a new app version, which is particularly useful given Google’s ongoing consolidation schedule.
Vertex AI teams should already be clear. Google placed the same Imagen 4 model IDs behind an earlier migration deadline of 30 June 2026 on Vertex AI. If you are running Gemini API workloads rather than Vertex workloads, the 17 August 2026 date applied to you.
No migration fee applies. There is no charge to migrate. Usage of gemini-3.1-flash-image is billed at Google’s standard image generation pricing. Compare your workload against the live pricing page before switching, as the cost per image may differ from what you were paying on the Imagen 4 endpoints.
One new capability worth knowing about
The gemini-3.1-flash-image model supports a video-to-image generation mode that was not available on the Imagen 4 endpoints. You can pass a video file, either via direct upload or a public YouTube URL, alongside a text prompt to generate thumbnails, cinematic posters, or summary infographics. This is exclusive to gemini-3.1-flash-image and not available on gemini-3-pro-image. If that is useful for your application, it is a straightforward addition once the migration is complete.
What is coming next
The retirements do not stop in August. Two more shutdowns are already on the schedule:
gemini-2.5-flash-imageshuts down on 2 October 2026- The Gemini 2.5 Pro, Flash, and Flash-Lite cluster shuts down on 16 October 2026
If you have any workloads on those model IDs, put those migration dates in your calendar now. The pattern Google has established is a deprecation notice of roughly two months before shutdown, so the window is not generous.
The Gemini API changelog and deprecations page are the authoritative sources for upcoming end-of-life dates. Both are worth bookmarking if you run anything on the Gemini API.