API quickstart

Python

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_RELIOVA_KEY",
    base_url="YOUR_ASSIGNED_BASE_URL"
)

response = client.chat.completions.create(
    model="YOUR_APPROVED_MODEL_ID",
    messages=[{"role": "user", "content": "Summarize this ticket."}],
    stream=True
)

for event in response:
    print(event.choices[0].delta.content or "", end="")

cURL

curl "YOUR_ASSIGNED_BASE_URL/chat/completions" \
  -H "Authorization: Bearer $RELIOVA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "YOUR_APPROVED_MODEL_ID",
    "messages": [{"role":"user","content":"Hello"}],
    "stream": true
  }'

Compatibility boundary

“OpenAI-compatible” describes the request pattern, not perfect feature parity. Supported parameters, tool calling, JSON behavior, thinking controls, usage accounting, context limits, and error responses are verified per model route and documented during onboarding.

Operational requirements

• Keep keys server-side and rotate them after staff or vendor changes.
• Set explicit timeouts, bounded retries, and idempotency where applicable.
• Log request IDs and usage fields without logging sensitive prompts by default.
• Pin an approved model ID; do not rely on an unversioned alias for regulated or high-risk workflows.