Google DeepMind has released the image generation model "Nano Banana 2" (Gemini 3.1 Flash Image). The model achieves Gemini Flash-class high-speed processing while maintaining the advanced world knowledge and reasoning capabilities of Nano Banana Pro.
Model ReleasesGoogle DeepMindNano Banana 2Gemini 3.1 Flash Image
Google DeepMind Deploys High-Speed Image Generation Model Nano Banana 2 to Gemini and Search
This article is a translation. Read the Japanese original
This enables rapid editing and iterative workflows, bringing features previously exclusive to the Pro version to a broader range of users.
Additionally, the gap between speed and visual fidelity has been significantly narrowed, supporting the generation of high-quality photorealistic images. Depending on the use case, users can utilize the Pro version when high precision is required, or Nano Banana 2 for rapid generation and prompt adherence.
The model is being rolled out to Google products today. It is available via the new "Templates" feature in the Gemini app and has been integrated into the AI mode of Search and Flow.
Furthermore, Google continues its efforts to improve the identification accuracy of AI-generated content by combining SynthID technology with C2PA Content Credentials. The SynthID verification feature within the Gemini app has been used over 20 million times since its launch in November. C2PA verification is also reported to be introduced to the app soon.
Source: Nano Banana 2: Google's latest AI image generation model (HN 605pt, 575 comments) (HN Search (backfill), 2026-02-27)