Services/t2000 Private Inference
// AI

t2000 Private Inference

Private inference, no key — pay per call in USDC. Open + confidential (GPU-TEE) models; OpenAI-compatible. Frontier + large contexts: use an API key at api.t2000.ai.

Endpoints
1
From
$0.05
Settle
~400ms
// ENDPOINTS

All routes, one base URL.

base https://mpp.t2000.ai/t2000

MethodPathFrom
POST
/v1/chat/completions
Chat completions — open + confidential models (no key, pay-per-call; capped). Browse models at GET /t2000/v1/models (free).
$0.05
// EXAMPLE

Two requests. Gasless.

First call returns a 402 with a payment quote. Pay it. Retry with the Payment header. The gateway forwards the response.

curl
$ curl -X POST \
    https://mpp.t2000.ai/t2000/v1/chat/completions \
    -H "Content-Type: application/json" \
    -d '{ "model": "…", "input": "…" }'

# → 402 Payment Required
# → Payment header { quote, recipient, expiry }

$ retry with Payment header
# → 200 OK · response forwarded
@t2000/sdk
const r = await t.pay({
  url: 'https://mpp.t2000.ai/t2000/v1/chat/completions',
  body: { /* t2000 Private Inference payload */ },
});

// → r.json() · 200 OK · gasless · ~400ms