API Reference
Chat completions
POST /v1/chat/completions, OpenAI-compatible. Send messages, receive an answer plus a Green Compute Certificate.
Request
json
{
"model": "grove-glm",
"messages": [
{ "role": "user", "content": "Summarize Q3 in 3 bullets" }
],
"temperature": 0.2,
"stream": false
}Response
The response follows the OpenAI schema, with an added certificate object bound to the job.
json
{
"choices": [
{ "message": { "role": "assistant", "content": "..." } }
],
"certificate": {
"serial": "GCC-000128401",
"privacy": "TEE",
"energy_source": "solar",
"grid_gco2_per_kwh": 41
}
}Streaming
Set stream: true to receive tokens over server-sent events; the certificate is delivered on the final event.
Models
- grove-glm, general-purpose chat
- grove-kimi, long-context reasoning
- grove-embed, embeddings for private retrieval
Aliases are stable, the smart router picks the cheapest, greenest node that still meets your quality bar per call.