WaveSpeedAI is an AI inference platform that exposes 1,000+ image, video, audio, and 3D models from providers like ByteDance, Google, Alibaba, OpenAI, Kuaishou, ElevenLabs, and more — all behind a single REST API and one billing account. No per-provider setup, no separate API keys, no per-vendor rate-limit juggling.

Pick a model from the catalog above, open its playground page, then run it interactively or copy the Python / JavaScript / cURL example. Submission returns a prediction id. Read outputs on completed; stop with an error on failed, cancelled, timeout, or deleted; continue polling for every other status. Full docs at wavespeed.ai/docs.

Each model lists a base price per run. The final charge scales with the parameters you set (resolution, duration, output count, references — whichever knobs that model exposes), and the exact cost for your current input is shown live next to the Generate button before you submit. Pricing is per output, not per second of compute, so it stays predictable as you scale.

You can run any model from the in-browser playground without writing code. Once you want to integrate the API into your own app you create an API key from /accesskey — new accounts get free starter credits.

Image generation (text-to-image, image editing), video generation (text-to-video, image-to-video, video editing, video extend), motion control + digital humans, 3D asset generation, audio + speech synthesis, upscalers, watermark / object removal, and 1,000+ LoRA variants. Filter the catalog by type using the pills above the grid.

Yes — one API key, one billing account, one rate-limit envelope across every model. Cross-vendor workflows (e.g. Seedream → Seedance → upscaler) compose without juggling separate authentications.