any-llm 1.0: Seamlessly Switch Between Cloud and Local LLMs with One Python Library

Mozilla.ai released any-llm v1.0, an open‑source Python library that provides a unified interface for communicating with different large language model (LLM) providers, allowing developers to switch between cloud and on‑premise models without rewriting code.

any‑llm 1.0, launched on November 4, is available on GitHub. According to Mozilla.ai machine‑learning engineer Nathan Brake, the library reduces boilerplate, eases integration, and gives developers the flexibility to choose any LLM regardless of vendor.

Version 1.0 includes a stable, consistent API, async‑first design, reusable client connections for high‑throughput and streaming use cases, clear deprecation and experimental feature notices, and a provider compatibility matrix.

Improved test coverage for stability and reliability

Response API support

Model‑listing API to query each provider’s supported models

Reusable client connections for better performance

Standardized inference output across all models

Automatic updates and a provider compatibility matrix

Future plans feature native batch completions, new provider support, and deeper integration with the any‑suite libraries such as any‑guardrail, any‑agent, and mcpd.

Relevant URLs: https://github.com/mozilla-ai/any-llm, https://github.com/mozilla-ai/any-guardrail, https://github.com/mozilla-ai/any-agent, https://github.com/mozilla-ai/mcpd.