Skip to content
Router One

One image generation API for every model

The same gateway key that reaches GPT, Claude, and Gemini for chat also generates images. Router One puts text-to-image models from multiple families — gpt-image-2, Gemini 3 Pro Image, Gemini 3.1 Flash Image, Doubao Seedream, Grok Imagine — behind one OpenAI-compatible endpoint at https://api.router.one/v1, with a flat per-image unit price and a cost trace for every request.

Text-to-image models in the catalog

The lineup follows the live catalog — each model's current per-image unit price is on the models page under the image category, and every detail page lists its supported endpoints with example calls.

gpt-image-2

OpenAI's current image model — strong prompt adherence and reliable text rendering inside generated images.

Gemini 3 Pro Image (Nano Banana Pro)

The Pro tier of Google's image line, known in the community as “Nano Banana Pro” — multimodal prompting with strong photorealism. It lists image input in the catalog, so it can take a reference image through POST /v1/images/edits.

Gemini 3.1 Flash Image (Nano Banana 2)

The Flash tier of the same line, the model the community calls “Nano Banana 2” — model ID gemini-3.1-flash-image-preview. Same request shape and the same per-image unit price as the Pro entry, and it also accepts a reference image on POST /v1/images/edits.

Doubao Seedream 5.0

ByteDance's Seedream series — notably strong Chinese-prompt understanding and CJK text layout.

Grok Imagine

xAI's image family, listed as two catalog entries — grok-imagine-image and grok-imagine-image-quality, the higher-fidelity pass.

Flat per-image pricing with a paper trail

Media models are unit-priced, so an image job costs one flat USD amount — and the gateway's observability applies to every generation.

  • One flat unit price per generated image, shown live in the catalog — no token accounting on media output.
  • Every request lands in Dashboard → Logs with model, unit cost, latency, and status — the same trace chat calls get.
  • Per-key budgets and spend caps apply, so a production image key is capped independently of your chat traffic.

Call it like the OpenAI Images API

POST /v1/images/generations with a model ID and a prompt — the request shape follows the OpenAI Images API, so existing OpenAI SDK image code works after changing the base URL and key. Copy a model ID from the catalog. Image-to-image edits go to POST /v1/images/edits instead: multipart/form-data, with the reference image uploaded as the image field.

image-generation.sh
# Text-to-image via the OpenAI-compatible endpoint
curl -X POST https://api.router.one/v1/images/generations \
  -H "Authorization: Bearer sk-your-router-one-key" \
  -H "Content-Type: application/json" \
  -d '{"model": "<image-model-id>",
       "prompt": "An orange kitten wearing an astronaut helmet"}'

Top up with a card, Alipay, or stablecoins

Wallet balance is metered pay-as-you-go: top up with a card or Alipay through one hosted checkout, or with USDT/USDC on six chains — no US credit card required.

Video generation on the same key

The catalog also lists video generation models — Vidu Q3 Turbo, PixVerse, MiniMax Hailuo 2.3, and Wan 2.6 (image-to-video) — served at POST /v1/videos/generations with per-clip unit pricing. Image and video share the same key, wallet, and cost traces.

No low fixed cap for normal paid usage. Abuse prevention, per-account protection limits, and upstream constraints may still apply; if a request returns 429, check Dashboard -> Logs or contact support to raise limits.

FAQ

Which image generation models are available?

The live lineup is on the models page filtered to the image category — currently spanning gpt-image-2, Gemini 3 Pro Image (Nano Banana Pro), Gemini 3.1 Flash Image (Nano Banana 2), Doubao Seedream, and Grok Imagine (grok-imagine-image and grok-imagine-image-quality). The catalog updates as new models ship.

How is image generation billed?

Per image, at a flat unit price shown in the catalog, deducted from your metered wallet balance like any other request. There is no token accounting on generated images, and each job's unit cost is recorded in its request trace.

Is the API OpenAI-compatible?

Yes — POST /v1/images/generations follows the OpenAI Images API request shape: set the base URL to https://api.router.one/v1, pass model and prompt. Any client that can call the OpenAI Images API works.

Can I use it from Mainland China?

Yes. The endpoint is reachable from Mainland China without a VPN, and you can top up in RMB with Alipay on the hosted checkout — no foreign card needed.

Does the same key also generate video?

Yes — video models such as Vidu Q3 Turbo, PixVerse, MiniMax Hailuo 2.3, and Wan 2.6 (image-to-video) are unit-priced per clip and served at /v1/videos/generations. Browse them in the catalog under the video category.

Choose the next step for your image-generation workflow

This page owns the first image-generation call. Use the pages below to select a current model, confirm the request contract, move to video, read the longer China guide, or fund the shared wallet.

Ready to generate images through one gateway key?

Get your API key