Skip to main content

OpenAI Compatibility

Kertex AI exposes an OpenAI-compatible API in addition to the native Gemini surface. Point any OpenAI client or tool at the environment's /v1 base URL, use your Kloudbean token as the API key, and it works — same models, same limits, same metering as the native endpoint.

Kertex AI OpenAI-compatible API

Base URL​

OpenAI clients call /chat/completions and /models under this URL. Use the full URL including /v1:

https://<environment-host>/v1
Example:  https://horizon-dev-kertexai.kloudbean.net/v1

This is different from the native Gemini base URL (which has no version) — OpenAI clients need the /v1. You can copy the exact URL from the OpenAI API tab in your console.

Authentication​

Use one of the environment's tokens as the API key. OpenAI clients send it as Authorization: Bearer <token>, which Kertex accepts natively. The Get token button on the OpenAI API tab takes you straight to token management.

Supported endpoints​

EndpointPurpose
POST /v1/chat/completionsChat completions (streaming and non-streaming).
GET /v1/modelsLists your tier aliases so clients can populate a model picker.

Pass a Kertex tier alias (kertexai-fast, kertexai-advanced, kertexai-light) as the model.

Using the OpenAI Python SDK​

pip install openai
from openai import OpenAI

client = OpenAI(
    base_url="https://horizon-dev-kertexai.kloudbean.net/v1",
    api_key="kb_dev_your_token_here",
)

resp = client.chat.completions.create(
    model="kertexai-fast",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Give me one tip for faster API responses."},
    ],
)
print(resp.choices[0].message.content)

Streaming works the same way:

stream = client.chat.completions.create(
    model="kertexai-fast",
    messages=[{"role": "user", "content": "Count to five."}],
    stream=True,
)
for chunk in stream:
    delta = chunk.choices[0].delta.content or ""
    print(delta, end="", flush=True)

curl​

curl "https://horizon-dev-kertexai.kloudbean.net/v1/chat/completions" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer kb_dev_your_token_here" \
  -d '{
        "model": "kertexai-fast",
        "messages": [{"role": "user", "content": "Hello!"}]
      }'

Connect from OpenWebUI​

  1. Open Admin Panel → Settings → Connections.
  2. Under OpenAI API, click + to add a connection.
  3. Set URL to the /v1 base URL above and Key to your Kertex token.
  4. Save — your model tiers appear in the model picker.

The same pattern works for LibreChat, Continue, LangChain, and any other OpenAI-compatible client: set the base URL to https://<environment-host>/v1 and the API key to your Kertex token.

What's preserved​

Requests through the OpenAI layer run the same pipeline as native calls — token authentication, spend caps, rate limits, and metering all apply identically. Only the wire format differs; Kertex translates between OpenAI and Gemini for you. In the Logs & History tab these calls appear with the action openai.chat.completions.

Next steps​