Nano Banana
Google Gemini 2.5 Flash Image (the original "Nano Banana") on a visual canvas. Fast image generation and edits, wired into publishable flows.
- Nano Banana is the community nickname for Google's Gemini 2.5 Flash Image model.
- Fast, affordable image generation and edit-by-description through the Gemini API.
- PlugNode exposes it as the default option on the Image node. BYO Gemini key, no markup.
- Upgrade path to Nano Banana Pro (Gemini 3 Pro Image) is a dropdown change.
What it is
Nano Banana is the nickname the generative-image community settled on for Google's Gemini 2.5 Flash Image model. It generates images from text prompts and accepts reference images with caption hints, positioned as a fast and affordable entry point into the Gemini image family. On PlugNode, it is the default model on the Image node. You set a prompt, optionally attach one or more reference images, pick an aspect ratio, and the node returns the generated image to downstream connections. The same Image node also lets you select Gemini 3.1 Flash Image or Gemini 3 Pro Image (Nano Banana Pro) from the dropdown when a flow needs more detail at the cost of latency and per-call price.
What you can do with it
- Text-to-image at Flash-tier latency and cost
- Image-to-image editing driven by natural-language instructions
- Multi-reference composition with caption hints per image
- Aspect-ratio control at config time
- Promote to Nano Banana Pro by changing the model dropdown
- Wire into HTTP Trigger and Respond to Webhook for a public API
Why the original Nano Banana still matters
Nano Banana Pro is stronger on detail and composition, but the original Nano Banana (Gemini 2.5 Flash Image) is still the right pick for two common cases.
First, high-volume generation: bulk catalog variants, A/B thumbnail sets, social cards at scale, where per-call latency and cost matter more than peak fidelity.
Second, flows where the image is one step in a longer pipeline and will be edited downstream anyway: generate a base, pass it into Image Resize, layer copy on top with a design step. You do not need peak fidelity at step one when step three will rewrite pixel-level detail.
Edit-by-description, the Flash way
Same edit-by-description contract as Nano Banana Pro: pass a reference image and a text instruction, get back an edited image. The output is less detailed than Pro, and the tradeoff is latency and cost.
On PlugNode a typical edit flow is: File Input, then Image (Nano Banana) with prompt, then Image Resize, then Respond to Webhook. Because the Image node config is the only thing that changes, switching to Pro is one dropdown and a re-publish. No other node in the flow cares which Gemini image model executed.
Multi-reference composition with caption hints
The Image node interleaves each reference image with a caption string before sending to Gemini. For Nano Banana, that means a prompt like "place the product" plus an image hinted "product on white cyclorama" followed by another image hinted "brand background plate" gives the model structured guidance rather than three floating references.
You can express this in PlugNode by passing an array of { image, text } items to the Image node. The executor preserves the order and wires the caption directly before its image in the Gemini request.
When to upgrade to Nano Banana Pro
Signals that Nano Banana is hitting its ceiling in a flow: customers complain the generated subject lacks detail or consistency across runs. Fine typography inside the image renders garbled. Compositing multiple references produces visible seams.
Each of those is a dropdown change on the Image node: switch to "Gemini 3 Pro Image (Preview)" and re-run. The rest of the flow, including the HTTP endpoint and version history, is untouched.
Nano Banana in a published API flow
A minimal Nano Banana API flow: HTTP Trigger accepts { "prompt": "..." }, Image (Nano Banana) generates, Respond to Webhook returns the image URL as JSON. Four nodes, one click to publish.
The resulting POST https://plugnode.ai/api/trigger/{secret}/{nodeId} is a drop-in image-generation endpoint your product can hit without bolting in the Gemini SDK. Rate limit is 60 requests per minute per trigger. SSRF protection is enabled by default on external URLs the flow references. Version history means you can roll back after a bad prompt edit without emergency hotfixes.
See the publish-as-API pillar for the full pattern, the Gemini Image integration for the node reference, and the HTTP Trigger integration for trigger-side details.
Run it on PlugNode
Nano Banana is the default option on the Image node, labelled "Gemini 2.5 Flash Image" in the UI. Supply a Gemini API key, wire inputs, run. Promote to Nano Banana Pro when detail matters by changing the dropdown.
Nano Banana generations are billed by Google at Gemini API rates. PlugNode adds no markup. Flash-tier pricing makes this the cheapest option in the Gemini image family.
Use it inside these workflows
All use casesFrequently asked questions
- What is Nano Banana?
- Nano Banana is the community nickname for Google's Gemini 2.5 Flash Image model. It is the fast, affordable tier of the Gemini image family and the default model on PlugNode's Image node.
- How is Nano Banana different from Nano Banana Pro?
- Pro uses Gemini 3 Pro Image and is larger, slower, and stronger on detail. The original Nano Banana uses Gemini 2.5 Flash Image, which is faster and cheaper at the cost of some fidelity.
- Can I edit images with Nano Banana?
- Yes. The Image node accepts a reference image and a text instruction and returns an edited image. Pair with the File Input node to accept uploads.
- Does PlugNode charge extra for Nano Banana calls?
- No. You pay Google directly at Gemini API rates with your own key. There is no PlugNode credit system on top.
- Can I publish a Nano Banana flow as an API?
- Yes. Add an HTTP Trigger node, end with Respond to Webhook, hit Publish. You get a signed, rate-limited URL with version rollback.
Related models
All modelsLast updated 2026-04-25
Generate your first video ad in 3 minutes.
Free to start. No credit card. Upload a product photo, connect your AI models, click Run.