Google Unveils 'Nano Banana 2': The Next Evolution of Gemini Image Generation

In late February 2026, Google officially launched Nano Banana 2 (technically designated as Gemini 3.1 Flash Image). This state-of-the-art model is now live, combining the breathtaking quality of previous Pro versions with the blazing speed of Google's lightweight Flash architecture. It replaces the original Nano Banana model as the default image generator within the Gemini App, while the heavier Nano Banana Pro remains accessible to paying subscribers via the "Redo with Pro" option.

Key Breakthroughs and Capabilities

Nano Banana 2 represents a massive leap forward in both basic generation and complex editing workflows:

• Dramatically Improved Text Rendering: Unlike older generators that struggled with spelling, Nano Banana 2 excels at rendering accurate text within images, making it an ideal tool for creating charts, infographics, advertising mockups, and greeting cards.
• Subject Consistency: The model has learned to reliably maintain the appearance of multiple characters and objects across a series of different generations. This is a critical upgrade for creators working on comic books, storyboards, or game art.
• World Knowledge: By tapping into Gemini's knowledge base and web search, the AI can generate images grounded in reality—such as creating a photorealistic view from a specific city window reflecting actual, current weather conditions.
• Advanced Editing (Image + Text-to-Image): Users can upload an existing image alongside a text prompt to perform precise, localized edits without altering the rest of the composition.
• Composition & Style Transfer (Multi-Image-to-Image): Users can input multiple images simultaneously, allowing the model to synthesize them into complex composite scenes or perfectly extract the artistic style from one image and apply it to another.

Broad Ecosystem Availability

Google is rolling out Nano Banana 2 across its core platforms, including Gemini Apps, Google Search (AI Overviews and Lens), and for developers via Google AI Studio and the Gemini API in Vertex AI.

The 'Redo with Pro' Feature and Quotas

While Nano Banana 2 serves as the lightning-fast default for everyone, Google ensures that premium users still have access to heavier, specialized compute when needed.

Subscribers on the AI Plus, Pro, and Ultra tiers utilize a unique workflow. To access the heavier "Nano Banana Pro" model, users first generate their initial concept using Nano Banana 2. If they require intense refinement, they simply click the three-dot menu on the generated image and select the "Redo with Pro" option.

However, this premium capability is bound by a combined daily image generation quota—currently up to 20 uses for the Basic tier, 50 uses per day for AI Plus, 100 for Pro, and 1,000 for Ultra subscribers, with limits subject to change over time. If a user exhausts this daily limit using the base Nano Banana 2 model, the Pro upgrade option remains locked until the quota resets.