Quickstart

1. Get a key

Sign up → Dashboard → API Keys → Create key → copy it (shown once).

export NIKCLI_API_KEY=nik_live_xxxxxxxxxxxxxxxxx

2. First request

curl https://inference.nikcli-ai.dev/v1/chat/completions \
  -H "Authorization: Bearer $NIKCLI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nikseek",
    "messages": [{"role":"user","content":"Hello"}]
  }'

The response is OpenAI-compatible. We add a nikcli field with cost breakdown, the upstream route, and cache status:

{
  "id": "...",
  "model": "nikseek",
  "choices": [{ "message": { "role": "assistant", "content": "Hi" } }],
  "usage": { "prompt_tokens": 10, "completion_tokens": 2 },
  "nikcli": {
    "provider": "openrouter",
    "upstreamModel": "deepseek/deepseek-v4-pro-20260423",
    "cache": "miss",
    "costUsd": 0.000164,
    "upstreamCostUsd": 0.0000131,
    "marginUsd": 0.000151,
    "rid": "..."
  }
}

3. Model selection

Pick by alias (recommended):

Append :thinking to enable chain-of-thought on hybrid models. See Reasoning.

4. Next

• Cursor → drop it in your editor
• OpenAI SDK →
• All models →