Google released Nano Banana 2.1, its latest image generation and editing model, on 6th October 2026, and the model is already rolling out in AI Mode in Google Search.
Nano Banana 2.1 Arrives
- Google released Nano Banana 2.1 on 6th October, 2026 as its newest model for making and editing images.
- Google says the model beats its earlier versions, with big gains in design, masked edits, and keeping subjects consistent.
- Robby Stein, Google’s VP of Product in Search, said the model is already rolling out in AI Mode.
- Search users open the Google app and tap the banana icon under the Search box to start creating.
- Google has brought Nano Banana to three Search features in about a year, though AI Overviews may still use an older version.
Google framed the release as a broad upgrade. In its announcement post on X, the company said Nano Banana 2.1 “outperforms our previous models across the board.” Google pointed to notable leaps in visual design, mask-based editing, and subject consistency. It said those changes help people create more natural-looking images.
Mask-based editing is the most practical item on that list. In image tools, a mask typically marks one region so an edit changes only that area. Subject consistency means the same person or object stays recognizable across edits.
Search received the model on launch day. Robby Stein, Google’s VP of Product in Search, posted on X that “Nano Banana 2.1 is rolling out in AI Mode in Search!” To try it, he said, users open the Google app and tap the banana icon under the Search box.
Meet Nano Banana 2.1, our latest image generation and editing model 🍌 This upgraded version outperforms our previous models across the board, with notable leaps in visual design, mask-based editing, and subject consistency to help you create more natural-looking images.
— Google (@Google) October 6, 2026
Search now gets Google’s image model on day one
Earlier versions reached Search in stages. Google first brought the original Nano Banana model to Google Lens about a year ago. It expanded access to AI Mode last December and added the model to AI Overviews in July.
The 2.1 release breaks that pattern. AI Mode carried the new model on the same day Google announced it. That timing suggests Google now treats Search as a primary venue for its wider Gemini AI push, with image creation sitting next to answers.
AI Overviews may still run an older model
The rollout looks uneven across Search. AI Overviews do not appear to use version 2.1 yet, while AI Mode does. Google’s posts mention AI Mode only and say nothing about AI Overviews or Lens.
The performance claims also come from Google alone. Its posts include no benchmark scores, no list of supported countries, and no schedule for other Search features. Google has not yet answered four questions:
- Which countries and languages get Nano Banana 2.1 in AI Mode first?
- When will AI Overviews and Lens move to version 2.1?
- Does the update reach AI Mode on desktop, or only the Google app?
- How does the “across the board” claim hold up in independent tests?
What’s Next?
People already making images in Search should note where they start. AI Mode output should come from 2.1 as the rollout reaches each account. AI Overviews results may still reflect the earlier model. Users who do not see the banana icon yet can update the Google app and check again later.
AI Overviews is the next marker to watch. Once Google moves that feature to 2.1, all three Search entry points for Nano Banana could share one model. Until then, the same image request may return different results depending on where a user types it.
