Google: Nano Banana (Gemini 2.5 Flash Image)

Google: Nano Banana (Gemini 2.5 Flash Image) is an image generation model from Google, identified on api14 as google/gemini-2.5-flash-image. It sits in the image category with text as a secondary category, and it was released by the provider on 2025-10-07. The model is generally available, and its knowledge cutoff is 2025-01-31.

The model accepts image and text as inputs and can produce both image and text as outputs, so a single request can mix a written prompt with reference images and return a generated image, text, or both. It has a context window of 32768 tokens and a maximum output of 8192 tokens. Image generation is controlled through dedicated image parameters: aspect_ratio accepts one of 1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9 or 21:9; n has a minimum of 1 and a maximum of 1, so each request returns a single image; and input_references accepts between 0 and 3 images. The model is built for contextual understanding alongside image generation, which is what allows reference images and text instructions to inform the result.

Practical uses follow from that mix of inputs and outputs. You can generate an image from a text prompt alone, or supply up to three reference images to guide style, subject or composition while describing the change you want in text. Because the model also returns text, it fits workflows where a written explanation or caption accompanies the generated image. The fixed set of aspect ratios makes it straightforward to target common formats such as square, portrait, landscape, wide and ultrawide outputs without post-processing crops.

You call the model through api14's OpenAI-compatible API using the model id google/gemini-2.5-flash-image. The supported request parameters are max_tokens, response_format, seed, stop, structured_outputs, temperature and top_p. Image specific settings such as aspect_ratio, n and input_references are passed alongside those parameters. Because the interface is OpenAI-compatible, existing clients and SDKs that already target that shape can point at api14 with the model id changed.

A few limits are worth noting when deciding. The context window is 32768 tokens and the maximum output is 8192 tokens, so very long prompts or large volumes of reference material may need to be trimmed. Each request returns exactly one image, so batch generation means multiple calls. The knowledge cutoff is 2025-01-31. If your task is purely text generation with no image component, a text focused model may be a better fit.