> ## Documentation Index
> Fetch the complete documentation index at: https://docs.yingtu.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Gemini 3.1 Flash Lite Image on YingTu

> Choose Nano Banana 2 Lite for current-generation 1K image generation and editing where throughput and cost control matter.

`gemini-3.1-flash-lite-image` is the 1K efficiency specialist in the current
Nano Banana family. It fits interactive products and high-volume jobs that do
not need 2K or 4K output.

Google describes the upstream model as a low-latency, cost-efficient generator
and editor optimized for 1K output. It supports text-and-image input and
interleaved image editing. See the official
[model page](https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite-image).

## What YingTu verified

Synchronous generation, asynchronous generation, and image editing all
completed at 1024 × 1024 with PNG output in the September 2, 2026 run. A `1K`
request is accepted; `2K` and `4K` are not available for this model on YingTu.

<Note>
  Google's upstream page targets sub-two-second latency. That is not a YingTu
  latency promise: the observed synchronous gateway request took 10.647 seconds
  in one run, including routing and generation.
</Note>

## Compare with Nano Banana 2

| Decision             | Flash Lite Image               | Flash Image                                            |
| -------------------- | ------------------------------ | ------------------------------------------------------ |
| YingTu 1K price      | USD 0.025                      | USD 0.025                                              |
| Higher resolutions   | No                             | 2K and 4K                                              |
| Upstream position    | Efficiency and high throughput | General-purpose quality, speed, and resolution balance |
| Verified 1K encoding | PNG in this run                | JPEG in this run                                       |

Choose Lite when the output will remain 1K and the newer efficiency-focused
model is the point. Choose Flash when the same integration may need 2K or 4K,
or when its broader upstream positioning better matches the job. Encoding is
observational, not a selection guarantee.

<CardGroup cols={2}>
  <Card title="Use Lite parameters" icon="sliders-horizontal" href="/en/models/image/nano-banana/gemini-3.1-flash-lite-image/parameters">
    Keep requests inside the 1K limits.
  </Card>

  <Card title="Compare the family" icon="columns-3" href="/en/models/image/nano-banana/comparison">
    Review Lite against the original, Flash, and Pro models.
  </Card>
</CardGroup>