Overview
- Gemini 2.5 Flash Image is rolling out today in the Gemini app and to developers via the Gemini API, Google AI Studio, and Vertex AI, with distribution partners OpenRouter and fal.ai broadening access.
- The model focuses on character consistency across edits, supports multi‑image fusion, handles targeted local changes via natural language, and leverages Gemini’s world knowledge in multi‑turn workflows.
- Google lists pricing at $30 per 1 million output tokens, with each image using about 1,290 tokens, which works out to approximately $0.039 per image.
- All outputs include a visible AI mark and an invisible SynthID watermark and metadata, with policy limits intended to curb misuse following earlier issues with image accuracy.
- The tool was previously tested anonymously as “nano‑banana” on LMArena, where it ranked at or near the top for image editing, and Adobe said it is making the model available in Firefly and Adobe Express.