llm_bridge v0.1.0

First public release of llm_bridge — a tiny, vendor-neutral LLM client library by Santander AI Lab.

Highlights

• One interface, pluggable providers. Write your app against LLMClient once and switch between OpenAI, AWS Bedrock, Google Gemini, a local OpenAI-compatible server, or your own backend — without touching your code.
• Zero required dependencies. The core (mock, callable) is pure standard library; each vendor SDK is an optional extra ([openai], [aws], [google], [all]).
• Bring your own backend. Wrap any function with the callable provider.
• Normalised responses. chat() / complete() return an LLMResponse with content, token usage and latency.

Install

pip install llm-bridge            # core only
pip install "llm-bridge[all]"     # + OpenAI, Bedrock, Gemini SDKs

Apache-2.0 licensed. Python 3.9+. See the README and CHANGELOG for details.