LeanLLM

Lightweight Python wrapper around LiteLLM with built-in usage tracking and label support.

Installation

pip install leanllm

Or install locally for development:

pip install -e .

Quickstart

from leanllm import LeanLLM

client = LeanLLM(api_key="sk-...")

response = client.chat(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Hello!"}],
    labels={"team": "backend", "feature": "onboarding"},
)

print(response.choices[0].message.content)

Labels

Every request accepts an optional labels dict. Labels are attached to the usage event logged for that call, making it easy to slice costs and latency by team, feature, environment, or any dimension you define.

Usage logs

Each call appends a JSON line to llm_logs.json (configurable via LEANLLM_LOG_FILE / LEANLLM_LOG_DIR env vars):

{"model": "gpt-4o-mini", "prompt_tokens": 12, "completion_tokens": 8, "total_tokens": 20, "latency_ms": 432.1, "labels": {"team": "backend"}}