Google's Imagen 4 Gets Revamped in Gemini API and AI Studio Preview - Here's What's New

Google introduced Imagen-4 and its higher-precision sibling, Imagen-4 Ultra, in a paid preview for Gemini API users and free testing in Google AI Studio. Google has already released many AI tools for users. But according to the tech giant, the new Imagen-4 model gets better with on-image typography and delivers sharper visuals than its predecessor, Imagen-3.

Both these versions of Imagen include synthetic-ID watermarks on all outputs. Imagen-4 charges $0.04 per image, while Ultra costs $0.06. Users requiring stricter prompt adherence can opt for the latter. Google plans to add more pricing tiers later and will soon allow developers to request higher usage limits.

The blog post offers sample images demonstrating different use cases, from retro postcards and cinematic fashion editorials to tech-heavy sci-fi comics, all created using the Ultra variant. These examples showcase stronger fidelity in handling fine details like small text and graphic overlays.

Other recent Google news –

• Google Rolls Out Gemini AI App to All Education Workspace Users For Free
• Google Slashes Smart TV Spend to Double Down on YouTube Ads
• Android Studio Gets Smarter, Faster, and More Cross-Platform at Google I/O 2025

Google integrated Imagen-4 models into Gemini API and AI Studio using a unified library alongside its multimodal Gemini models . Developers in supported regions can generate images directly through API calls or experiment in AI Studio. All generated images will carry SynthID watermarks to ensure provenance and trusted usage .

Earlier this week, Gemini API release notes confirmed the June 24 launch of both Imagen-4 and 4 Ultra preview models. This aligns with Google’s recent strategy of rapidly rolling out multimodal tools, such as Veo 3 and Lyria 2, on platforms like Vertex AI during its May I/O keynote.

By expanding the Gemini API library, Google gives developers direct access to cutting-edge image generation capabilities. The pricing structure reflects a common cloud-AI pattern: standard models serve general needs, and higher-tier models target use cases requiring precise output. Imagen-4’s enhancement in typography responsiveness opens possibilities in design, marketing, and education workflows.

The Gemini API preview is now live through existing accounts, letting developers integrate Imagen-4 models into their applications immediately. Google anticipates broader access and additional tiers soon, as model availability and pricing evolve over the coming weeks.

You may also be interested to read –

• Microsoft’s 2025 Responsible AI Report Shows Deeper Governance and Global Reach
• Character.AI Hires Meta’s Former Biz Exec to Lead the Chatbot Company into Its Next Phase
• Meta Considered Buying Perplexity Before Betting on Scale AI