Documentation

FAQ

Short answers to the questions that come up first.

General

Do I need a separate account with each provider?

No. One Sothe key reaches every model in the catalog, and one balance pays for all of them.

Does my existing OpenAI code work?

Yes, once you change the base URL, the key and the model id. A few parameters are ignored — they are listed under Chat completions.

Is there a free tier?

No. You pay for what you use, from a balance you top up. See Pricing.

Where do I see what I spent?

Logs lists every request; Activity charts it by day, model and key.

Money

Why was I refused when I still had credits?

Because a request reserves the most it could cost before it runs. A large max_tokens on an expensive model can reserve more than your balance even if the answer would have been short. Lower max_tokens or top up.

Do credits expire?

No.

What is the 4% fee?

It is added to a top-up, not to your requests, and covers card processing. $5 of credits costs $5.20.

Not built yet

Things people ask for that genuinely do not exist, so you do not go looking:

  • Structured outputs — response_format is accepted but not enforced.
  • Automatic routing and fallbacks between models.
  • Embeddings, image generation, speech and transcription endpoints. Only chat completions exists.
  • Bring-your-own provider keys.
  • A management API, so management keys currently do nothing.
  • Idempotency keys.

The endpoints that do exist are POST https://api.sothe.net/v1/chat/completions and GET https://api.sothe.net/v1/models.