What is Google Pics and who can actually use it today?

Google Pics is a browser-based tool that lets you generate pictures from text prompts and then edit specific parts of the result without starting over. You reach it by typing pics.new into the address bar.

The announcement landed on September 1, 2026. Meg Whittenberger, the product manager for Google Pics, wrote the post on the official Google Products blog.

Who gets access first

The rollout targets two groups. Google AI Pro and Ultra subscribers see it first. Most Workspace business customers are in the same wave. The post does not give a hard date for personal Google accounts or for Workspace tiers outside the "most business customers" phrasing.

Access expands over the coming weeks. That is the only timeline the announcement provides.

What the tool actually does

Nano Banana powers the generation side. You type a description and the model returns several variants at once. The announcement calls this "multiple generations" and says you pick the one you like best.

Editing happens inside the same window. Object segmentation lets you click an element — a person, a logo, a background shape — and change only that piece. You can type a comment on the selected area and request a targeted edit. Several such edits can be queued and executed together.

Text that lives inside the image is editable in place. You can rewrite a headline on a poster or translate a label on a product mockup without disturbing the surrounding design. The font and layout stay intact.

Collaboration works like a shared Doc. You send a link, teammates open the same canvas, and everyone can add or tweak elements in real time.

Where it lives inside Workspace

Integration starts today with Docs and Slides. Drive follows in the coming weeks. The idea is to keep you inside the document you are already editing. You insert an image, realise the chart needs a different colour, and fix it without opening a new tab.

Drag-and-drop from Drive into Pics is mentioned as a way to pull existing assets into a new design. The post shows a short video of someone dropping a product photo into the canvas and building a brand concept around it.

What the announcement does not say

Pricing for the Pro and Ultra tiers is not repeated here. The Nano Banana model card — parameters, training data, benchmark scores — is not linked. No comparison to Imagen, DALL·E, or Midjourney appears. Offline use, mobile apps, and API access are not mentioned. How many images a subscriber can generate per month is absent.

The post calls the feature experimental in the AI-generated summary block at the top. The main body does not repeat that label.