Overview
- The release is live for free and paid users in the Gemini app and available to developers via the Gemini API (preview), Google AI Studio, and Vertex AI, priced at $30 per 1 million output tokens (about $0.039 per image).
- Key capabilities include prompt-driven local edits, multi-image fusion, multi-turn refinements, and use of Gemini’s world knowledge, with a focus on preserving character likeness across edits.
- All created or edited images carry Google’s invisible SynthID watermark, with visible “ai” markings appearing in some product surfaces to help label AI-made content.
- The model first appeared anonymously as “nano-banana” on LMArena, where it rose to the top of image-edit rankings; early tests from outlets report strong fidelity with some edge-case limitations.
- Distribution extends through partners OpenRouter and fal.ai, and Business Insider reports Adobe will make the model available in Firefly and Adobe Express.