One gateway, explicit governance
Authenticate with gateway keys, apply plan and budget policy, and keep the request path consistent for your application.
EU control plane for BYOK AI
Built for European software teams that need one OpenAI-compatible endpoint, enforced project budgets, understandable diagnostics, and provider billing that stays on their own accounts.
const client = new OpenAI({
apiKey: process.env.TENSORPROXY_API_KEY,
baseURL: process.env.TENSORPROXY_BASE_URL,
});
await client.chat.completions.create({
model: "gpt-4.1-mini",
messages,
});What the product actually does
Authenticate with gateway keys, apply plan and budget policy, and keep the request path consistent for your application.
Customer traffic still bills against the customer's own provider account rather than a pooled platform key.
Team plans can define provider order so requests keep moving when a preferred model path is unavailable.
Project budgets are enforced on the request path instead of being decorative dashboard numbers.
Review usage, request duration, token counts, and fallback reasons per request without exposing prompts or secrets.
Keep gateway configuration, team controls, budgets, and diagnostics on European infrastructure. Upstream inference location remains determined by the provider account and endpoint you choose.
How teams work through it
Add your OpenAI or Anthropic key once, then keep that boundary separate from your application runtime credentials.
Give each environment a scoped key and direct existing SDK traffic at the OpenAI-compatible endpoint.
Watch request health, review activity, and control fallback order without changing the application contract again.
Start with the real product surface
No card required. Create an account, add a provider key, and send your first request through the gateway.