Nano Banana 2
Google DeepMind · USA / UK · 2026
Google's default image model: 4K generation at Flash speed and roughly half the price of the Pro tier.
Nano Banana 2 (model ID gemini-3.1-flash-image) is the general-purpose image generation and editing model of the Gemini 3.1 family, announced on February 26, 2026. Google positions it as the default choice for most image work: it brings 4K output and the fidelity previously reserved for the Pro tier down to Flash-class speed and cost. It generates at 512 px, 1K, 2K and 4K, accepts up to 14 reference images for blending and style transfer, renders legible stylised text, and can be steered through conversational multi-turn editing rather than one-shot prompting. A distinctive addition is video-to-image: the model can take a video as context — including a YouTube link — and produce stills or derived artwork from it. Like the Pro tier it grounds generations in Google Search when a picture has to reflect current facts, and it runs a reasoning pass over composition before drawing. Pricing is per generated image and scales with resolution, from about USD 0.045 at 512 px to USD 0.151 at 4K, with batch processing at half those rates. Every output carries an invisible SynthID watermark. Access is through the Gemini API, Google AI Studio and Vertex AI.
▸Videos
No videos yet.
▸Reviews
No reviews yet. Be the first!