Google’s own image tool, Google Pics, has switched its core model. Recently, Google released an official Workspace update announcement, announcing that Google Pics is being upgraded to Nano Banana 2.1, improving the quality of image generation and editing. This upgrade addresses the things users complain about most: generated images aren’t detailed enough, image edits tend to alter parts that shouldn’t be touched, and text layout in images is messy. The official solution is to swap out the underlying model; the features themselves have not been added or removed.

What was changed?
The official announcement lists four changes:
Image quality and realismIt can produce high-resolution details, more natural lighting, and more accurate material textures when generating and adjusting images.
Editing Accuracy and Consistency: Whether it is a local edit or an overall adjustment, it will be more faithful to the original image and greatly reduce the impact on surrounding people, background, or brand elements.No changes were requested.This is the direct solution to the “image edits going off track” problem.
Text renderingWhen generating new images or editing text within existing images, character shapes and layout come out cleaner and more accurate. This is especially noticeable for Chinese users, since image models tend to blur strokes together or distort them when handling non-Latin scripts.
speedQuality has improved while processing speed remains similar to the previous version, with no slowdown.

Besides the standalone Google Pics app, users who call up Pics to edit images in Google Docs and Google Slides will also get the features from this update. This matters more than it sounds. When creating presentation visuals or document illustrations in Docs and Slides, images are often part of the output, and any misstep in editing an image will go straight into your final product. In the past, this kind of work mostly required putting the image into another tool, fixing it, and pasting it back; now it can be done in the same workflow.
Which plans can be used?
The officially listed scope of application is as follows:
The Business edition covers Business Standard and Business Plus; the Enterprise edition covers Enterprise Standard and Enterprise Plus; individual subscriptions cover Google AI Plus, Google AI Pro, and Google AI Ultra; the education add-on is Google AI Pro for Education; the other add-on is AI Expanded Access.
This isn’t an update just for enterprise customers—Google AI’s individual paid plans are on the list too. But note that on the Gemini side, Nano Banana 2.1 is still a feature of the paid Google AI plans; the free tier uses Nano Banana 2.
What is Nano Banana 2.1?
First, some background. Nano Banana is Google’s image generation and editing model, part of the Gemini family, and 2.1 is the current latest version. When Google announced Nano Banana 2.1 on October 7, its official statement was that it outperforms the previous model across the board, with the most notable improvements in three areas: visual design, mask-based image editing, and subject consistency.
Comparing these three items with this Workspace announcement shows that both are talking about the same set of issues. Mask-based editing corresponds to “local edits that stay on track,” subject consistency corresponds to “preserving the original image content,” and visual design corresponds to “image quality and materials.” When improvements at the model layer reach the product, they come down to those few things in users’ hands.
The Google Pics tool itself launched on Google Workspace in May 2026, focusing on professional-grade AI image generation and editing. It can also be invoked directly in Docs and Slides, so the benefits of this model swap extend beyond standalone apps.
What’s the actual difference?
Putting the official statements into scenarios users will actually encounter, they broadly come down to these. Previously, when generating portraits, skin texture or clothing fabric often had a plastic feel; this time the official announcement specifically calls out “material texture” as needing to be more accurate. Previously, if you only said “change the background to an office,” the pattern of the clothes on the person in front would change too; now the official word is that such unrequested changes will be greatly reduced. Previously, when you put a piece of text in an image, Chinese character shapes often came out blurry or skewed; this time text rendering is listed as a separate improvement item. On speed, the official wording is “similar to the previous version,” which amounts to a clear statement that it will not get slower just because the model has changed; for people who slot Pics into their workflow, this is a practical confirmation.

Two things to note
First, this is one timeModel Replacement, not a feature update. The interface, options, and workflow are unchanged, so you don’t need to adjust your existing habits; but it also means that if you have any opinions about the old version’s output style, there will be nothing else to adjust this time.
Second, the admin side.No option to choose the model version.The announcement states that there is no separate admin toggle; organizations that want to keep the old model do not have this option and can only disable Google Pics entirely.
For general users, the key consideration with this round of updates is simple: if your Google AI plan or company Workspace account is on the eligibility list, just wait for it to roll out to you—no configuration is required. What to watch are use cases where outputs go directly into final deliverables, such as images for presentations, illustrations in documents, and visual assets for social posts. The three most noticeable improvements this time—image quality, edits that stay on target, and text within images—are exactly what these areas care about most.
And if you don’t have a paid plan, you’ll still be using Nano Banana 2 on Gemini—this upgrade doesn’t extend to free users.
Source: KOCPC Chinese