unclecode-litellm

What it is and what it does

unclecode-litellm is a pre-compromise fork that abstracts away differences between LLM provider APIs, letting you call models from many providers through a single OpenAI-compatible interface. It ships as both a Python SDK for direct integration into your code and an AI Gateway (proxy server) for centralized LLM access with authentication, virtual keys, cost tracking, and request routing.

The Python SDK handles completion calls, embeddings, image generation, audio transcription, and more across providers. The AI Gateway layer adds multi-tenant support, per-project customization (logging, guardrails, caching), and admin dashboards for monitoring. It also supports agent-to-agent (A2A) protocols, MCP (Model Context Protocol) tool bridging, and router-based retry and fallback logic across multiple deployments. Runtime dependencies include httpx, openai, pydantic, aiohttp, and tokenizers.

Use it for:

• Switch between LLM providers without rewriting completion calls by changing the model string.
• Build a centralized LLM gateway for your organization with virtual keys, cost tracking per user/project, and unified authentication.
• Implement retry and fallback logic across multiple LLM deployments using the Router for resilience.
• Connect MCP servers to any LLM via the experimental MCP bridge for tool-augmented completions.
• Track LLM spend and usage across teams using the proxy's multi-tenant cost tracking and observability callbacks.
• Invoke agent-to-agent (A2A) protocols from multiple agent platforms through a unified gateway.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Unified Python interface to call many LLM providers (OpenAI, Anthropic, Bedrock, Azure, VertexAI, Groq, etc.) in OpenAI format, plus an AI Gateway proxy server for centralized access with authentication, cost tracking, and routing.

Yes, with conditions. This is a pre-compromise fork, so verify that this fork's divergence from the main project meets your needs—check whether all providers and features you require are present and functional. Install friction is low, maintenance is active, and the MIT license is permissive. No known vulnerabilities. Suitable for teams building multi-provider LLM applications or operating a centralized LLM gateway.

Install

unclecode-litellm on PyPI

pip install unclecode-litellm

uv add unclecode-litellm

poetry add unclecode-litellm

Installing unclecode-litellm

Before you install

Low install friction with a pure-Python wheel distribution. Active maintenance with recent commits. Twelve runtime dependencies (httpx, openai, pydantic, jinja2, aiohttp, click, tiktoken, tokenizers, jsonschema, importlib-metadata, python-dotenv, fastuuid) are all standard ecosystem libraries.

License in practice

MIT license permits unrestricted use, modification, and distribution in commercial and private projects with minimal obligations—only attribution required.

Quickstart

pip install unclecode-litellm

from litellm import completion
import os

os.environ["OPENAI_API_KEY"] = "your-key"
response = completion(
    model="openai/gpt-4o",
    messages=[{"role": "user", "content": "Hello!"}]
)

Requires Python 3.9 or later. API keys for the LLM providers you intend to use must be set as environment variables.

Verify before relying

• How this fork diverges from the upstream project and whether all advertised providers are fully functional.
• Whether all endpoints listed (/chat/completions, /embeddings, /images, /audio, /batches, /rerank, /a2a, /messages) are production-ready or experimental.
• Actual performance characteristics under production load beyond claims in the description.

Package facts

Evidence: unclecode_litellm-1.81.13-py3-none-any.whl

Tags

More Artificial Intelligence packages