alaada — Orbit AI Python SDK

Official Python client for the Orbit AI API by Alaada.

pip install alaada-orbit

Quickstart

import alaada as orbit

orbit.configure(api_key="orbit_live_...")

# Single prompt
res = orbit.prompt("simplex1", "What is the capital of France?")
print(res.text)

# Streaming
for chunk in orbit.stream("simplex1", "Write a short poem about the ocean"):
    print(chunk.delta, end="", flush=True)

# Multi-turn chat (history stored automatically)
chat = orbit.chat("simplex1", system="You are a Python coding assistant.")
r1 = chat.send("What is a decorator?")
r2 = chat.send("Show me an example.")   # full context retained
print(r2.text)

Models

API reference

orbit.configure()

orbit.configure(
    api_key="orbit_live_...",   # or set ORBIT_API_KEY env var
    privacy=False,              # True = server logs no prompt/response content
    default_model="simplex1",
    timeout=120,
)

orbit.prompt(model, prompt)

res = orbit.prompt(
    "simplex1",
    "Summarise this article: ...",
    system      = "You are a helpful assistant.",
    tools       = ["websearch"],        # optional
    temperature = 0.7,
    max_tokens  = 1000,
    privacy     = True,                 # override per-call
    session_id  = "chat_abc123",        # attach to a session
)

print(res.text)        # the response text
print(res.usage)       # {"input_tokens": 12, "output_tokens": 40}
print(res.image_url)   # set when using imagery1

orbit.stream(model, prompt)

for chunk in orbit.stream("simplex1", "Tell me a joke"):
    print(chunk.delta, end="", flush=True)
    if chunk.finish_reason == "stop":
        print(f"\n\nTokens used: {chunk.usage}")

orbit.chat(model)

session = orbit.chat(
    "complex1",
    system        = "You are an expert data scientist.",
    short_history = True,    # only keep the last chat_max exchanges
    chat_max      = 20,
    privacy       = True,
)

# .send() — full response
res = session.send("Explain gradient descent")
print(res.text)

# .stream() — streamed response
for chunk in session.stream("Now show Python code for it"):
    print(chunk.delta, end="", flush=True)

print(f"\nSession has {session.turns} turns")
session.clear()   # delete session from server

Tools

# Available tools: "websearch", "calculator", "datetime"
res = orbit.prompt(
    "simplex1",
    "What is the latest news on AI regulation?",
    tools=["websearch"],
)

Privacy

When privacy=True, the Orbit server will never log your prompt or response content — only routing metadata (method, status, duration) is recorded.

orbit.configure(api_key="...", privacy=True)   # global
res = orbit.prompt("simplex1", "...", privacy=True)  # per-call

Environment variables

Error handling

from alaada import AuthenticationError, RateLimitError, ValidationError, APIError

try:
    res = orbit.prompt("simplex1", "Hello!")
except AuthenticationError:
    print("Invalid API key")
except RateLimitError as e:
    print(f"Rate limited. Resets at {e.reset_at}")
except ValidationError as e:
    print(f"Bad request: {e}")
except APIError as e:
    print(f"Server error: {e}")

License

MIT — © Alaada
All rights reserved by the company and Arzo Das.