Google DeepMind released Nano Banana 2.1. It is a multimodal model in the Gemini 3 series. It accepts text and images as input. It outputs images up to 4K tokens and text up to 64K tokens. It uses a 1M token context window. It is based on the Gemini 3.6 Flash architecture. It is available in the Gemini App, Google AI Studio, and via API. It scores 1050 Elo in overall text-to-image preference. It scores 1048 Elo in infographic design. It shows improved factuality in infographics. It supports multi-turn editing and design tasks. It retains known limitations like small text rendering and occasional spatial confusion. It has a knowledge cutoff of March 2026.
What changed
Nano Banana 2 used a smaller context window and lower output limits.
What it unlocks
Generating and editing images with higher precision and longer text outputs in one model.
- 1M token context window
- 4K token image output
- 64K token text output
- 1050 Elo overall preference (Thinking)
What you need to act on it
- access via Gemini App, Google AI Studio, or API
Sources