Inference, with the border built in.
Sigmix Inference API is one API with two routes. The sovereign route is served only inside the Kingdom; the international route reaches frontier models abroad. We run the gateway, the billing and the GPUs on our own side of it.
One API, two routes
The OpenAI request shapes for chat, embeddings, speech, transcription and images, on two base paths. Your existing client works by changing the base URL and the key.
Our GPUs and named partners
Sovereign requests run on Sigmix GPUs in the Kingdom or on in-Kingdom partners. International requests run on named providers abroad or on Sigmix GPUs abroad. The gateway that chooses between them runs in Riyadh.
Residency per route
We do not make one residency claim for the whole service, because one would be false. The sovereign route stays in the Kingdom and is enforced at the key, the path and the database. The international route may leave it, and says so.
Counted by us, to the micro-dollar
Every request passes through our gateway, so we reserve before it runs and settle on the model's own count when it finishes. The console shows each request as it lands.
No sample figures presented as measurements
Every model, price and response time on this site is read live. Where there is no data, the page says so instead of showing an example.