Genorbex SDK for Python

The official Python client for Genorbex Cloud. It provides named clients for the platform's service families, universal authenticated REST access, GFlow Prompt Registry integration, and OpenAI-compatible model serving.

Installation

python -m pip install genorbex-sdk

The distribution exposes the modern genorbex_sdk import and retains genorbex and gflow for notebook compatibility.

Authentication

export GENORBEX_API_KEY="your-api-key"
export GENORBEX_HOST="https://www.genorbex.com"

Managed Genorbex notebooks receive short-lived credentials automatically.

Cloud services

from genorbex_sdk import GenorbexClient

client = GenorbexClient()

warehouses = client.database.list("warehouses")
instances = client.compute.list("instances")
pipelines = client.data_pipelines.list()
notebooks = client.data_engineering.list("notebooks")

Named clients cover cloud services, compute, object storage, Drive, OrbexDB, DBVisual analytics, data pipelines, data engineering, workspace, ingestion, migration, AI/ML, agents, streaming, connectors, orchestration, identity, teams, notifications, sustainability, web services, domains, security, and integrations.

Every named service supports list, get, create, update, replace, delete, and the lower-level request method. The top-level universal request method makes every current or future Genorbex endpoint accessible:

result = client.request(
    "POST",
    "/api/orbexdb/query",
    data={"sql": "SELECT 1", "warehouseId": "warehouse-id"},
)

GFlow prompts and model serving

import gflow
from genorbex_sdk import GenorbexClient

prompt = gflow.genai.load_prompt("prompts:/main.data_science.sales_summary@production")
formatted = prompt.format(
    number_of_sentences=3,
    sales_data="Revenue increased by 12% in July.",
)

client = GenorbexClient().serving_endpoints.get_open_ai_client()
response = client.chat.completions.create(
    model="genorbex-gpt-5-4",
    messages=[{"role": "user", "content": formatted}],
)
print(response.choices[0].message.content)

Python 3.9 or newer is required. The runtime uses only the Python standard library.

License

MIT