- Google has officially announced “Nano Banana 2 (Gemini 3.1 Flash Image),” the latest image generation AI model in the “Nano Banana” series based on the Google AI “Gemini.”
- Utilizing the “Flash” architecture of the “Gemini 3.1” generation, this latest image generation AI model achieves incredible speeds.
- It is being rolled out to the web version of the “Gemini” app for “Google AI Pro” and “Google AI Ultra” members.
On Thursday, February 26, 2026, Google officially announced “Nano Banana 2 (Gemini 3.1 Flash Image),” the latest image generation AI model in the “Nano Banana” series based on the Google AI “Gemini.”
“Nano Banana 2” is a cutting-edge image generation AI model that combines the best of both “Nano Banana” (announced in August 2025) and “Nano Banana Pro” (announced in November 2025) to deliver advanced global knowledge, high quality, and reasoning at incredible speeds.
“Nano Banana 2”
“Nano Banana 2 (Gemini 3.1 Flash Image)” is the latest image generation AI model that utilizes the “Flash” architecture of the “Gemini 3.1” generation to achieve astonishing generation speeds.
While maintaining the complex reasoning capabilities and high image quality of “Nano Banana Pro,” the image generation time has been significantly reduced. This allows users to fine-tune AI images through dialogue with “Gemini” in a stress-free environment that feels almost real-time.
Additionally, its reasoning capabilities through language understanding have been enhanced. It can now faithfully reproduce long-form prompts that specify complex spatial arrangements, such as “a red car in the back left and a cat holding a coffee cup in the foreground.”
The accurate text rendering feature, which was highly acclaimed in “Nano Banana” and “Nano Banana Pro,” has been further strengthened in “Nano Banana 2.” It can now precisely render user-specified text onto objects within the image, such as signs, neon lights, and documents. Furthermore, users can re-edit generated images using natural conversation, such as “change the color here to blue” or “make the background a city at night.”
Features such as creating new compositions by combining multiple reference images or extracting a specific artistic style from one image to apply it to another now operate more seamlessly. Naturally, a digital watermark called “SynthID” is embedded into the generated images to certify them as AI-generated content.
Notably, “Nano Banana 2” also includes measures against deepfakes, with restrictions in place for generating or editing images of politicians and real-life celebrities.
“Nano Banana 2” is being rolled out to developers as the “Gemini API” via the multimodal generative AI development platform “Google AI Studio” and the machine learning development platform “Vertex AI.” It is also available via the web version of the “Gemini” app for “Google AI Pro” and “Google AI Ultra” members under the “Google One” AI plans. A new style selection feature is also available in “Nano Banana 2.”
Please note, however, that there is a usage limit of up to 1,000 times per day.
Source：Google
コメントを残す