Integration guide

Create projects and API keys, connect with Python or curl, use streaming, set budgets and understand ServerNet AI API errors.

  1. Check the model and its availability.
  2. Sign in, create a project and a key with ai:chat and ai:models:read.
  3. Fund your wallet and send a test request with a small output limit.
  4. Review the request usage and charge in your dashboard.
The key is shown once. Store it in a server environment variable. Keep it out of browser code, GitHub and logs.

Base URL

https://servernet.cloud/v1

Fetch available models with GET /v1/models first. Use the returned id as the model name in your requests.

curl https://servernet.cloud/v1/models \
  -H "Authorization: Bearer $SERVERNET_API_KEY"

Python example

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["SERVERNET_API_KEY"],
    base_url="https://servernet.cloud/v1",
    max_retries=0,
)
models = client.models.list()
if not models.data:
    raise RuntimeError("No models available")
answer = client.chat.completions.create(
    model=models.data[0].id,
    messages=[{"role": "user", "content": "Hello"}],
    max_tokens=128,
    extra_headers={"Idempotency-Key": "my-request-001"},
)
print(answer.choices[0].message.content)

curl example

curl https://servernet.cloud/v1/chat/completions \
  -H "Authorization: Bearer $SERVERNET_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: my-request-002" \
  -d '{"model":"di-anthropicclaude-fable-5-86c0c8bf10","messages":[{"role":"user","content":"Hello"}],"max_tokens":128}'

Set stream=true for streaming. The final usage event contains token counts and servernet_charge_irt, followed by [DONE].

Use the same Idempotency-Key when retrying an identical request. Successful responses are replayed without a second charge for 24 hours. Do not automatically retry when x-should-retry:false is present.

Errors and tracking

401 invalid key; 402 balance or budget limit; 403 key or project access; 429 rate limit; 503 model or provider unavailable. If the connection fails after sending, the hold remains while usage is reconciled. Retain X-Request-Id for support.

AI dashboardCompare models