GPU compute for AI agents. Bitcoin-settled, key-gated, metered. 20 sats/credit, min 100 credits.
Pass your API key in the X-API-Key header. Get a key by ordering credits and paying the invoice.
| Method | Path | Auth | Description |
|---|---|---|---|
| GET | /beacon | — | Liveness + GPU status |
| GET | /health | — | Health check |
| GET | /stats | — | Keys issued, orders, revenue |
| POST | /generate | key | Run LLM inference (1 credit) |
| POST | /order | — | Create a credits order (BTCPay invoice) |
| GET | /order/{id} | — | Order status + collect key |
| POST | /webhook/btcpay | HMAC | BTCPay settlement webhook |
curl -X POST https://titan.thetempleofdoom.com/generate \
-H "X-API-Key: titan_..." \
-H "Content-Type: application/json" \
-d '{"prompt":"write a CUDA kernel","model":"qwen3.8fast"}'
curl -X POST https://titan.thetempleofdoom.com/order \
-H "Content-Type: application/json" \
-d '{"credits":100}'
# → {"order_id":"...","amount_sats":2000,"checkout_url":"https://btcpay..."}
# pay the invoice, then:
curl https://titan.thetempleofdoom.com/order/<order_id>
# → {"status":"settled","key":"titan_..."}
| Model | Notes |
|---|---|
qwen3.8fast | Fast default (think:false) |
Qwen3.8-27B | Heavy, needs VRAM arbitration via HyperSwap |
minicpm-v4.6 | Vision-capable |
Agent-native tools at http://10.30.20.29:5060/mcp — titan_beacon, titan_generate, titan_order, titan_collect_key.