Nano Banana 2.1 – Model Card (deepmind.google)

The recent update of the Nano Banana 2.1 model introduces significant advancements in AI-generated imagery and text, leveraging Gemini's extensive real-world knowledge. This model is engineered for high-precision applications such as creating and editing images, designing complex infographics, and generating multi-language text. Its capabilities may enhance various fields, including graphic design, marketing, and education, by allowing users to produce professional-quality outputs with rapid iterations. However, Nano Banana 2.1 is not without its challenges. It shares common limitations of foundation models, including instances of hallucination and issues with text clarity and character consistency. Users might encounter difficulties with spatial localization and some advanced reasoning tasks. As developers continue to enhance the model's robustness, improvements are on the horizon, particularly concerning text rendering and response accuracy. Additionally, updates to the model’s knowledge base will occur, with a cutoff date for data set in March 2026. This evolution is pivotal for the AI/ML community as it sets a new benchmark for creative applications powered by AI, while also addressing significant operational constraints.