Quickstart

From zero to your first /v1/chat/completions call in five minutes.

1. Create an account

2. Mint an API key

Go to /dashboard/keys and click New API key. You'll get one of two prefixes:

cx_live_… — production traffic, real billing.
cx_test_… — non-billable test calls (good for CI).

The plaintext key is shown once. Store it in your secrets manager immediately — Conduix only stores a hash.

3. Point your SDK at Conduix

Conduix is OpenAI-API-compatible. Change two lines in your existing client and you're routing.

from openai import OpenAI

client = OpenAI(
    api_key="cx_live_YOUR_KEY",
    base_url="https://api.conduix.ai/v1",
)

response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Hello, Conduix!"}],
)
print(response.choices[0].message.content)

4. Read the response headers

Every response includes observability headers — most useful is x-conduix-provider, which tells you which upstream actually served the request (especially valuable when a fallback kicked in).

bash

x-conduix-provider:     openai
x-conduix-model-served: gpt-4o-mini
x-conduix-latency-ms:   842
x-conduix-cache:        MISS
x-conduix-request-id:   cx-req-abc123…

See Observability for the full list and how they're used.

5. Watch your usage

Open /dashboard/usage to see token counts, request counts, and per-model spend in real time. Top up credits at /dashboard/billing before you run out — or enable auto-reload.

What's next

Authentication — how API keys work, when to rotate, scoping with allowlists
Models & pricing — the catalog, fallback chains, and per-1M-token rates
Governance — spend caps, region pinning, PII redaction

Authentication