SeedRouter is a unified AI model API that puts image, video, language and audio models behind a single key, letting developers send one request and pay only for delivered results. It is built for teams that want to call many different models without managing separate provider accounts, SDKs or billing relationships.
What is SeedRouter?
SeedRouter is an API gateway that exposes image, video, language and audio models from providers including OpenAI, Anthropic, Google, ByteDance, Moonshot AI and DeepSeek through one key. You send a request to a chosen model and receive the generated output — text, image, video or audio — while paying per request or per unit of output. Failed requests cost nothing, and there is no monthly minimum.
What makes SeedRouter stand out?
- One key for every model — image, video, language and audio models from multiple providers are reachable through a single API key.
- Pay only for what is delivered — billing is per request or per second/image/token, and a request that fails costs nothing.
- No monthly minimum — there is no subscription floor; you pay as you go.
- OpenAI and Anthropic SDK compatibility — models such as GPT-6 Astra, Kimi K3, DeepSeek V4.1 Flash and Claude Fable 5.1 can be called with the OpenAI or Anthropic SDK.
- Live per-token and per-second pricing — language models show per-1M-input-token rates, video models show per-second rates, and image models show per-image rates.
- Playgrounds and docs per model — every model card links to parameters, a playground and documentation.
- Large context windows — several language models offer 1M to 1.05M-token context.
- Video generation range — Seedance 2.0 family models output 480p to 4K, 4–15 seconds, while Seedance 2.5 supports up to 30 seconds at 1080p.
Who is it for?
- Application developers — integrate text, image, video and audio generation into a product without wiring up multiple provider accounts.
- AI product teams — compare and switch between models such as GPT-6 Luna, Claude Opus 5.5 or Nano Banana 2 using one billing relationship.
- Video and image creators building tools — call Seedance 2.0/2.5 or Nano Banana models for generation and editing workflows.
- Prototypers and startups — test models in the playground and pay per request with no monthly minimum.
What can you do with SeedRouter?
- Text and reasoning: Call GPT-6 Astra, GPT-6 Sol, GPT-6 Luna, Kimi K3, DeepSeek V4.1 Flash, Claude Fable 5.1, Claude Fable 5 or Claude Opus 5.5 with the OpenAI or Anthropic SDK.
- Video generation: Generate video with Seedance 2.0, 2.0 Fast, 2.0 Mini or 2.5 using text, first and last frames, and image, video or audio references.
- Image creation and editing: Use Nano Banana 2, Nano Banana Pro, Nano Banana 2 Lite, GPT Image 2.5 or GPT Image 2 with text prompts, reference images and transparent backgrounds.
- Model evaluation: Open any model card for parameters, a playground and docs before committing to an integration.
Pricing
SeedRouter uses pay-as-you-go pricing with no monthly minimum, billed per request or per unit of output. Language models are priced per 1M input tokens (for example, GPT-6 Luna from $0.08, DeepSeek V4.1 Flash from $0.24, GPT-6 Sol from $1.60, Kimi K3 from $2.40, Claude Opus 5.5 from $3.20, and GPT-6 Astra, Claude Fable 5.1 and Claude Fable 5 from $8.00). Video models are priced per second (Seedance 2.0 Mini from $0.0182, Seedance 2.0 Fast from $0.0548, Seedance 2.0 from $0.0909, Seedance 2.5 from $0.1329). Image models are priced per image (GPT Image 2 from $0.00425, GPT Image 2.5 from $0.00448, Nano Banana 2 Lite from $0.02565, Nano Banana 2 and Nano Banana Pro from $0.03).
FAQ
Is SeedRouter free?
No. SeedRouter is a pay-as-you-go API: you are billed per request or per unit of output, such as per 1M input tokens, per second of video or per image. There is no monthly minimum, and a request that fails costs nothing.
Which SDKs can I u
