Gemini 2.5 Flash Image (Nano Banana)

2025-08-26 · Publisher : Google · Domain : Artificial Intelligence

Google launches Gemini 2.5 Flash Image, nicknamed Nano Banana, its image editing and generation model: image fusion, character consistency, targeted edits in natural language and world knowledge.

Google launches Gemini 2.5 Flash Image, nicknamed Nano Banana, its image editing and generation model: image fusion, character consistency, targeted edits in natural language and world knowledge. ### What changes - Fusion of several images into one, keeping a character consistent from one scene to the next. - Precise edits in natural language: blur the background, remove a stain or a person, change a pose, colorize a photo. - Uses Gemini's world knowledge for semantic uses (diagrams, guided edits). - Available from the announcement in the Gemini API, AI Studio and Vertex AI; stable version on October 2, 2025. - All images carry an invisible SynthID watermark. ### Specifications - Identifiers: `gemini-2.5-flash-image-preview`, then `gemini-2.5-flash-image` (GA). - Inputs: image and text; input context of 65,536 tokens and output of 32,768 tokens; knowledge cutoff: June 2025 (API documentation). - Distribution partners: OpenRouter and fal.ai. The model is announced as deprecated, with shutdown planned on October 2, 2026 according to the current pricing page. ### Pricing | Item | Price | |---|---| | Image output | 30 USD per million tokens (1,290 tokens per image, i.e. 0.039 USD per image) | | Other modalities | Gemini 2.5 Flash pricing | On November 4, 2025, the number of input tokens per image went from 1,290 to 258, which reduces the cost of edits. ### Benchmarks No figures in the blog post. ### Key takeaways The model that popularized conversational image editing. Google replaced it with successors from 2026: plan the migration.

Tags : Image, Google, Gemini, Nano Banana, Flash