SeedRouter
SeedRouter brings image, video, language, and audio models behind one API: a single key, a single balance, pay-per-use billing, and no charge for failed requests.
About SeedRouter
SeedRouter is a model API aggregation gateway: users only need to create one API key and top up one balance to call image, video, language, and audio models with the same request format, including GPT, Claude, Gemini, GPT Image, and video and image model families such as Seedance, Flux, and Qwen Image. Billing is per request, per image, per video second, or per token, and charges apply only for successful output—failed requests are free. The model list gets updated, but the calling method stays the same. For developers who don't want to register separate accounts, manage multiple keys, and juggle multiple bills for every model provider, it consolidates multiple suppliers into a single integration point. Video and some image and audio models run as async jobs: the request returns a job id, which you can poll or attach a webhook to, and the result is retrieved via a CDN link once complete.
Best for
Best use cases
Add text, image, and video generation capabilities in a single integration
Add AI features to an existing product, starting with a small trial before scaling up
Retrieve long-running generation results such as video via async jobs and webhooks
Manage usage and spend across multiple model providers in one place
Features
Single API key for multiple model types
One key to call image, video, language, and audio models, with all models using the same request format and billing method.
Usage-based billing, no charge for failures
Failed requests are not charged; you pay only for successful output. No subscription, no monthly minimum, and the balance never expires.
Covers mainstream model families
Supports GPT, Claude, Gemini, GPT Image, as well as Seedance, Flux, Qwen Image, and more; list updates require no changes to existing calls.
Async jobs and webhooks
Time-consuming jobs like video return a job id that can be polled or paired with a webhook, with results delivered via a CDN link.
Image editing capabilities
Image models like GPT Image 2 support text prompts, reference images, and masked editing, all called through the same endpoint.
Pros
- One key and one balance to call image, video, language, and audio models, replacing per-provider sign-ups and invoicing
- Usage-based billing with no subscription and no monthly minimum spend
- Failed requests are not charged, and the balance never expires
- Model list updates require no changes to existing calls
- New accounts get $0.10 in free credit to try it out first
Cons
- Async APIs such as image models do not support stream or partial_images, and are not a replacement for the synchronous Images SDK (as stated in the documentation)
- Requests are rejected on insufficient balance (402) or too-frequent requests (429), requiring a top-up first or backoff retries (per the documentation's error code table)
Pricing Plans
Pay-as-you-go top-up
- Every USD 1 topped up becomes USD 1 in balance, usable on any model
- Balance never expires, with no monthly reset
- Billed at each model's published unit price
- Failed requests are not charged
* Pricing information is for reference only. Please refer to the latest announcement on SeedRouter's official website.