Loading the catalogue…
Loading the catalogue…
Start in minutes. Scale to enterprise with the same API. Inference is billed at cost-plus on top — and cost-aware routing means the easy 80% stops paying frontier prices.
Move the sliders — your numbers, not ours. See routed cost vs a flat frontier plan, and how many vendors you collapse into one.
Illustrative only: assumes a sovereign request ≈ €0.002 and a frontier request ≈ €0.02. Your actual token volumes and rates will differ — this estimates the routing effect, not a quote.
A flat platform fee per tier, plus transparent cost-plus on the tokens you actually use. The Smart Router sends routine work to cheap sovereign models, so your effective rate is far below a flat frontier subscription.
Yes. Attach your own keys for any catalogue provider and requests route through your account — your pricing, your limits — with a small flat platform fee and a per-tier free allowance.
No. GPT-5.5, Claude Opus 4.6 and Gemini 3.1 Pro are one route away on Professional and above — governed by your sovereignty policy and logged like everything else.
No. Prompts and completions are never used to train models. Sovereign requests stay under EU jurisdiction; routed requests are governed and logged with full provenance.
One line. Point your existing OpenAI SDK at api.stav.ai/v1 and keep your code.
Sovereign by default, frontier on demand, and a bill that matches the work.