SeedRouter

One API for image, video, language and audio models

SeedRouter introduction

SeedRouter puts image, video, language and audio models behind a single API key with pay-per-request billing. Requests that fail cost nothing, and long-running jobs return a task id that can be polled or delivered by webhook.

Website:: seedrouter.ai

🌄 Art & Image Generatorpaid

SeedRouter overview

SeedRouter is a platform layer that provides one API in front of image, video, language and audio models. According to the source, users create one key, load a balance in USD and call any supported model with the same request shape and the same billing, replacing separate accounts, keys and invoices for each model vendor. Language models are served through an OpenAI-compatible endpoint, while images, video and audio use the SeedRouter task API. Pricing is listed per model — per request, per image, per second of video or per token — and billing occurs when the output is delivered. Failed requests are not charged. Video and some image and audio jobs run as tasks: the request returns a task id, which can be polled or registered with a webhook, and output arrives as a CDN link. A dashboard tracks usage and cost per request across every model, with balance, keys and top-ups in USD. The source states there is no subscription and no monthly minimum.

SeedRouter features

One API key for image, video, language and audio models
Same request shape across model families
OpenAI-compatible endpoint for language models
SeedRouter task API for images, video and audio
Pay-per-request billing charged on delivery
Failed requests are not charged
No subscription and no monthly minimum
Task id returned for video and long image jobs
Polling or webhooks for asynchronous jobs
Output delivered as a CDN link
Dashboard for balance, keys and usage
Cost per request visible across every model

Questions about SeedRouter

SeedRouter pros

One key and one billing account across image, video, language and audio models
Failed requests are not charged
No subscription and no monthly minimum
Language models work with the OpenAI SDK by changing the base URL
Same request shape when switching between model families
Webhooks and task polling for long-running video and image jobs
Per-model prices are listed before you call
Dashboard shows usage and cost per request across every model

SeedRouter limitations

Pricing is usage-based per request, image, second or token, so cost scales with volume
Video and some image and audio jobs are asynchronous and require polling or a webhook
A model priced in tiers is listed at its standard rate, and a longer request can settle at a higher tier
When a model is unavailable the request returns an error instead of a result

SeedRouter use cases

Generate and edit video from text or a first frame
Create images from text or edit with reference images
Build chat, reasoning, code generation and agent workflows
Generate songs, voices and dialogue from a prompt
Separate stems from an existing track
Switch between GPT, Claude and Gemini model families
Move an existing OpenAI SDK integration by changing the base URL
Run long video and image jobs asynchronously with webhooks

Who SeedRouter is for

Developers integrating AI models into products
Product teams building production AI features
Engineering teams that want one billing account across model vendors
Teams using the OpenAI SDK for language models
Builders needing image, video, language and audio models from one key

SeedRouter pricing

Pay per request against a USD balance, with no subscription and no monthly minimum. Each model has a listed price — per request, per image, per second of video or per token — and you are charged when the output is delivered. Failed requests cost nothing. Examples listed include GPT Image 2 from $0.00425 per image, Nano Banana 2 from $0.03 per image, Seedance 2.5 from $0.1080 per second, and Claude Opus 5.5 from $3.20 per 1M input tokens. A model priced in tiers is listed at its standard rate, and a longer request can settle at a higher tier.