Google's Gemini API deprecation schedule lists all three Imagen 4 model IDs with the same shutdown date: August 17, 2026. The recommended replacement for every one of them is gemini-3.1-flash-image.

The facts: the three affected IDs are imagen-4.0-generate-001, imagen-4.0-ultra-generate-001 and imagen-4.0-fast-generate-001. Two other entries sit in the same window — gemini-robotics-er-1.6-preview shuts down on August 31, 2026, with gemini-robotics-er-2-preview as its replacement. Google's page notes that the dates published are the earliest announced shutdown dates, with exact timing communicated separately, so August 17 is the date to plan against rather than a date to test your luck on.

Automation Squad's take: model deprecations are the least glamorous and most disruptive item in any AI stack, because the thing that breaks is never the thing you are currently looking at. It is the n8n workflow you built in March that generates a product image once a night, or the Make scenario behind a client's social posts. Those fail silently into an error branch nobody reads. The swap itself is genuinely one line — it is finding every place that line lives that takes the afternoon. The good news is that the replacement is a Gemini-family image model rather than a third-party migration, so the request shape barely changes.

Run this now: search your automation tools and codebase for the string "imagen-4.0" — in n8n, that means checking every HTTP Request and Google Gemini node; in Make or Zapier, every custom API call; in code, a plain project-wide find. For each hit, swap the model ID to gemini-3.1-flash-image and run the workflow once manually to confirm the output still lands where it should. Do the test run before August 17, not after, because a failed run after the shutdown looks identical to a broken credential and you will waste an hour chasing the wrong thing.