Imagen 4
Google DeepMind · USA · 2025
Google's last stand-alone image generator: three tiers from two to six cents an image, 2K output and legible type - retired on 17 August 2026 after barely a year, as Google folds image generation into Gemini.
Imagen 4 is the fourth and final generation of Google's dedicated text-to-image family, announced at Google I/O in May 2025, released in preview on 24 June 2025 and made generally available on 14 August 2025. It shipped in three tiers that differ in price and ceiling rather than in character: Fast at two cents an image for drafts and bulk work, Standard at four cents, and Ultra at six cents for prompts that must be followed to the letter. Standard and Ultra render up to 2K (2048x2048, or 2816x1536 in the widest ratio); Fast stops at 1K. All three take a text prompt in English, Chinese, Hindi, Japanese, Korean, Portuguese or Spanish, return up to four images per request in five aspect ratios, offer configurable safety settings and person-generation controls, and stamp every output with a SynthID watermark that Google's own tooling can verify. The reason to read this profile is the ending rather than the specification. Google is not replacing Imagen with Imagen 5 - it is dissolving the product line into Gemini, where images are generated by the same model that handles text through the ordinary generate_content call. That leaves developers with two shutdown dates rather than one, which is a genuine trap: on Vertex AI the Imagen 4 endpoints were discontinued on 30 June 2026, while on the Gemini API the same models run until 17 August 2026. Migration is not a matter of swapping a model name, because the dedicated generate_images method disappears with the family. What also disappears is the price floor: Imagen 4 Fast at two cents an image was the cheapest first-party image generation Google ever sold.
▸News
▸Videos
No videos yet.
▸Reviews
No reviews yet. Be the first!