Nano Banana 2 Lite Is Live in Momen: 1K Image Generation at a Quarter of the Cost

Nano Banana 2 Lite — listed in Momen as Gemini-3.1-flash-lite-image — is now in Momen's model list. You select it inside an AI agent and wire that agent into your app's backend logic. No API key, no provider account.

The reason to care is cost per image. A 1K image on this model costs exactly half of Gemini-3.1-flash-image-preview and exactly a quarter of Gemini-3-pro-image-preview — roughly five cents at top-up rates, worked out in full below.

Understanding Gemini-3.1-flash-lite-image

This is the fast, cheap member of the Nano Banana family — the same generation and editing behavior as its siblings, tuned for volume instead of maximum fidelity. It generates an image in as little as four seconds, with edits running slightly slower than fresh generations. That speed is the point: it makes generated images viable in places you would not have put them before, like a user waiting on a screen rather than a background job.

Three limits shape what you build with it.

1K is the ceiling. 1024×1024 is the only size it produces — there is no 2K or 4K variant to switch to. For higher resolution, use Gemini-3.1-flash-image-preview (available at 0.5K, 2K and 4K) or Gemini-3-pro-image-preview (2K and 4K).

Output is plain text and images, not JSON. Structured output is unavailable, so an agent on this model cannot return a schema you parse downstream. Streaming stays available.

The agent cannot call tools. It will not reach for an Actionflow, an API, or another agent on its own. Give it a prompt, take back an image.

Together those make it a generation step rather than a reasoning agent. When you need JSON or tool calls around the image, run a text model in a second agent and chain the two in an Actionflow.

How to Use Gemini-3.1-flash-lite-image in Momen

Momen connects models two ways: built-in models that run on Momen's provider accounts and bill in AI Points, and BYOM, which runs calls on a provider account you own.

Built-in model

Gemini-3.1-flash-lite-image is built in, on every plan including Free. Nothing to connect, no key to store, usage billed in AI Points.

1. Open the AI tab in the header of your Momen project.
2. On the Agent tab, click Add agent (or open an existing one).
3. Click the model name at the top of the agent editor to open its settings.
4. Click the Model: row and pick Gemini-3.1-flash-lite-image from the list.
5. Configure the rest of the agent: temperature, max number of rounds, maximum output tokens, and image processing (Simple or Detailed).
6. Drop the agent into an Actionflow to connect its output to a database write, a file upload, or an API call.

BYOM (Bring Your Own Model)

BYOM runs the same model through your own provider account, so calls bill to that account rather than to your AI Points. It requires Basic or above.

1. Open the AI tab, then the Model tab.
2. Click Add model and connect the provider account you want the calls to run on.

For how agents connect to the rest of your backend, see the AI integration guide. For a worked image-generation build, see Build a Product Image Generator.

What It Costs

Point conversion

Momen bills AI usage in AI Points, converted from tokens at a rate that differs per model. Every model's rate is listed in the Point conversion table on the AI page, next to your AI Point balance.

What one image costs

The formula is the same for every model:

points = (input tokens × input rate) + (output tokens × output rate)

A 1K image is billed as 1,120 output tokens. So:

1,120 × 25.71 ≈ 28,795 points per generated image.

The prompt is rounding error next to that. A 30-token text prompt adds 30 × 0.43 ≈ 13 points. What does matter is reference images: each one you pass in is billed as 1,120 input tokens, so 1,120 × 0.43 ≈ 482 points per reference image. A text-to-image call costs ~28,800 points; an edit with two reference images costs ~29,760.

Against the other two Gemini image models, at the same 1K size:

Half the cost of the flash preview model, a quarter of the pro one — the token count is identical at 1K, and the output rates are exactly 2× and 4×. Push those two models to 2K or 4K and they bill more tokens per image, widening the gap further.

How far the Free plan goes

The Free plan includes 100,000 AI Points per month.

100,000 ÷ 28,795 ≈ 3.5 — so three images per month, and the fourth one runs you out.

That is a plan for trying the model, not for running on it. The same arithmetic for the paid plans:

Anything with real volume runs on the AI Points add-on rather than on the included allotment.

Plans and top-ups

AI Points top up at 6,000,000 points for $10/month. That makes one point worth $10 ÷ 6,000,000, so:

28,795 × ($10 ÷ 6,000,000) ≈ $0.048 per image.

Roughly five cents an image, or about 208 images per $10 block. On the same basis Gemini-3.1-flash-image-preview is ≈$0.096 and Gemini-3-pro-image-preview is ≈$0.192.

Plans start at Free, with Basic at $39/project/month and Pro at $99/project/month (less on annual billing). Basic is the first plan with unlimited AI agents; Free is capped at one. To size your own numbers against storage, traffic and Actionflow runs as well as points, use the pricing calculator.

Real Showcases

Apps already built on Momen that run on image generation — the same slot Gemini-3.1-flash-lite-image drops into.

Building a Production-Ready AI Image Generator with Nano Banana and No-Code A full image generator running on an earlier Nano Banana model — the same build, one model swap away. Read more.

Conclusion

A model on its own is not a product. What makes image generation shippable is everything around the call — a database for what gets generated, an Actionflow that fires on submit, file storage, accounts, payments — and that is what Momen gives you. Start from the app you want to build, whether that is a paid generator or avatars in a product you already run, and let this agent be one step in the flow.