kronaxis-router 0.1.0

Python SDK for Kronaxis Router - intelligent LLM proxy

Kronaxis Router Python SDK

Zero-dependency Python client for Kronaxis Router.

pip install kronaxis-router

Quick Start

from kronaxis_router import KronaxisRouter, Tier

router = KronaxisRouter("http://localhost:8050", service="my-app")

# Simple chat (auto-classified tier)
response = router.chat("What is the capital of France?")

# Force tier 2 (cheap model) for extraction
data = router.chat(
    "Extract the email from: Contact us at [e-mail]",
    tier=Tier.LIGHT,
    call_type="extraction",
)

# Force tier 1 (powerful model) for reasoning
plan = router.chat(
    "Design a 3-phase migration strategy for...",
    tier=Tier.HEAVY,
    max_tokens=2000,
)

# LoRA adapter routing
email = router.chat(
    "Draft a cold outreach email...",
    model="sdr",  # Routes to backend with SDR adapter
    system="You are a sales development representative.",
)

# Batch submit (50% off)
job = router.batch_submit(
    backend="cloud-fast",
    requests=[
        {"custom_id": "1", "body": {"model": "default", "messages": [{"role": "user", "content": "..."}], "max_tokens": 100}},
        {"custom_id": "2", "body": {"model": "default", "messages": [{"role": "user", "content": "..."}], "max_tokens": 100}},
    ],
    callback_url="https://my-app.com/webhook",
)
print(f"Batch job: {job['id']}")

# Check costs
costs = router.costs()
print(f"Today's spend: ${costs['daily'].get('my-app', 0):.4f}")

API

• router.chat(prompt, *, system, model, max_tokens, temperature, tier, priority, call_type)
• router.chat_messages(messages, *, model, max_tokens, temperature, tier, priority, call_type)
• router.batch_submit(requests, backend, callback_url)
• router.batch_status(job_id)
• router.batch_results(job_id)
• router.costs(period)
• router.health()