Cloud in three steps
- Register at /app/register and confirm Free (or complete paid checkout).
- Create an API key on /app — keys look like zk_ followed by hex.
- Point the OpenAI SDK at https://zenithgw.com/api/v1 and call model auto.
cloud_quickstart.py
# pip install openai
from openai import OpenAI
client = OpenAI(
base_url="https://zenithgw.com/api/v1",
api_key="zk_YOUR_CLOUD_KEY",
)
print(client.models.list())
print(client.chat.completions.create(
model="auto",
messages=[{"role": "user", "content": "ping"}],
))Enterprise in five steps
- Purchase Enterprise and download the .lic certificate from Cloud /app/licenses.
- Install the Gateway with install.sh, or follow Docker / Swarm for manual Compose/Swarm.
- Activate at /admin/activate (issuer must be reachable; /readyz is 503 until licensed).
- Configure a provider, create a pool alias, and note the bootstrap or created API key.
- Point the OpenAI SDK at https://your-domain.demo/v1 with model set to your pool alias.
install-zenith.sh
curl -fsSL https://zenithgw.com/install.sh -o install-zenith.sh
less install-zenith.sh
bash install-zenith.sh --yes
# Production pin example:
# bash install-zenith.sh --yes --tag ghcr.io/vestavision/zenith-gateway:v2026.07.26.0Verify the contract
Interactive API references on the Cloud origin: /docs/api/cloud (hosted chat-first) and /docs/api/gateway (full Gateway /v1 surface, including embeddings, audio, images, MCP). On a self-hosted Gateway origin, Scalar remains at /docs with OpenAPI at /openapi.json. HTTP smoke without an SDK: scripts/openai-conformance-smoke.sh (set OPENAI_API_PREFIX=/api/v1 for Cloud).