Nano Banana Series
by Google DeepMind
Google’s image generation model family built on Gemini Flash and Gemini Pro — fast, high-quality image synthesis with character consistency and text rendering
See https://gemini.google/overview/image-generation/
Model family
| Version | Release | Model ID | Notes |
|---|---|---|---|
| Nano Banana | Aug 2025 | Gemini 2.5 Flash Image | Original; fast image gen with experimental video frames |
| Nano Banana Pro | Nov 2025 | Gemini 3 Pro Image | Production-grade 4K; 14 reference images; ~$0.24/image at 4K |
| Nano Banana 2 | Feb 2026 | gemini-3.1-flash-image-preview | ★ Current default; #1 Artificial Analysis arena; ~$0.067/2K image |
Key capabilities (Nano Banana 2 — current)
- Speed — 4–6 seconds typical at 2K
- Resolution — 512px to 4K, 14 aspect ratios
- Character consistency — up to 5 characters across iterative edits
- Text rendering — precision multilingual text in images
- Real-time search grounding — integrates Google Search for data-accurate visuals (unique to v2)
- Configurable reasoning — minimal / high / dynamic thinking modes
- SynthID watermarking — C2PA provenance on all outputs
Access
- Gemini app — 🍌 icon or “Create an image…”; default for all users from v2
- Google AI Studio — developer access, full parameter control
- Google Flow — batch generation with zero credits
- Gemini API / Vertex AI — programmatic access
Pricing (Nano Banana 2)
- ~$0.067 per 2K image
- ~$0.039–0.134 per image depending on resolution
- Approximately half the cost of Nano Banana Pro
Related notes
nano-banana-pro.md— Nano Banana Pro detail notenano-banana-2.md— Nano Banana 2 full detail note (model specs, prompting guide, use cases)google-flow-nano-banana.md— Google Flow + Nano Banana workflow (in ai-coding/)