# Nano Banana — Independent API and Local-Model Guide

> Nano Banana 2 is Google's cloud-hosted Gemini 3.1 Flash Image model. This independent site explains how to call the API from Python and compares it with separate open-weight models that can be self-hosted.

## Identity and scope

- Site: https://www.nano-banana.live/
- Status: independent community resource; not an official Google or Gemini website
- Nano Banana 2 model ID: `gemini-3.1-flash-image`
- Execution: Google hosts the model. A Python script can run locally, but inference does not.
- Local alternatives: separate open-weight projects with their own licenses and hardware requirements

## Start here

- [Cloud API vs local models](https://www.nano-banana.live/)
- [Route chooser and model guide](https://www.nano-banana.live/guides/quickstart.html)
- [Python image-editing example](https://www.nano-banana.live/blog/nano-banana-api-image-editing-python.html)
- [Qwen-Image-Edit-2511 local guide](https://www.nano-banana.live/blog/qwen-image-local-image-editing-guide.html)
- [HunyuanImage 3.0 hardware reality](https://www.nano-banana.live/blog/hunyuanimage-local-deployment-requirements.html)
- [Benchmarks](https://www.nano-banana.live/benchmarks/)
- [Prompt recipes](https://www.nano-banana.live/prompts/)
- [FAQ](https://www.nano-banana.live/faq.html)

## Verified model distinctions

### Google cloud models

- Nano Banana 2: Gemini 3.1 Flash Image (`gemini-3.1-flash-image`), cloud API only
- Nano Banana 2 Lite: Gemini 3.1 Flash Lite Image, cloud API only
- Nano Banana Pro: Gemini 3 Pro Image, cloud API only

Official source: https://ai.google.dev/gemini-api/docs/image-generation

### Open-weight alternatives

- Qwen-Image-Edit-2511: Apache-2.0 image-editing model; official Diffusers example is on its model card
- HunyuanImage 3.0: 80B-total/13B-active MoE. The official repository recommends at least 3 × 80 GB VRAM for the base model and 8 × 80 GB for Instruct variants
- HiDream-O1-Image: 8B model under the MIT license

Official sources:

- https://huggingface.co/Qwen/Qwen-Image-Edit-2511
- https://github.com/Tencent-Hunyuan/HunyuanImage-3.0
- https://github.com/HiDream-ai/HiDream-O1-Image

## Evidence policy

- Prefer provider documentation, official repositories, model cards, and dated benchmark snapshots.
- Do not treat a local API client as locally running inference.
- Do not repeat rankings, prices, VRAM figures, or release status without a date and source.
- Generated blog drafts require human review before indexing or publication.

Sitemap: https://www.nano-banana.live/sitemap.xml