Powered by a generative AI API

A generative AI API lets software create new media — images, video, audio — by sending a prompt to a hosted model over HTTP. You don't run GPUs, you don't manage model weights; you send a request, poll the job, and get back a finished asset. For a product like TuneSizzle, that means one song idea can fan out into lyrics, audio, artwork and video without us operating any inference hardware ourselves.

generative AI API (generativeaiapi.com) is what we use for the visual side of that pipeline. It exposes a pool of image and video models behind one REST endpoint, all priced in credits so the cost of every generation is known.

• One endpoint for image and video — the same request shape drives covers, seed images and reels.
• Credit-based pricing per model, so the cost of every generation is known before it runs.
• A pool of models at different price and quality tiers — drafts to flagship renders — matching our own model pool.
• Queued jobs with stored outputs returned as signed URLs, which we then copy into durable storage.
• Automatic refunds when a generation fails — failed jobs never eat into anyone's credits.

Because the API speaks in models and credits, TuneSizzle's Admin → Models pool maps directly onto it: we can add new image or video models, price them per block, and let you pick them per song or per block in the video editor.

If you're building something of your own, the same API is available directly. A single POST with a model slug and a prompt starts a generation; poll the job or pass wait: true to get the result inline. Scoped API keys, per-key usage attribution and automatic failure refunds are built in. See the docs and current pricing at generative AI API at generativeaiapi.com.