SnapGen

SnapGen is one OpenAI-compatible key for the text, image, and video models in the live catalog. Rates are shown before you call, and usage comes off a prepaid balance instead of a subscription. Happy to answer questions about the API.

A prepaid balance across GPT/Gemini/Seedance instead of juggling four vendor accounts is a real pain point. Curious how you handle rate-limit failover if one upstream model goes down mid-request — does SnapGen retry on a fallback model automatically, or does the caller have to handle that?

An OpenAI-compatible gateway is a practical idea because switching only needs a new base URL and key. I also like that rates are shown per token and per call, which makes cost planning much easier. Two questions: how do you handle changes in the live catalog, such as a model being retired or a price change? And is there a way to set a spending cap on the prepaid balance?