Overview
llm-chat-completions-server is an LLM plugin developed by Simon Willison, designed to start a local server compatible with the OpenAI Chat Completions API, thereby exposing all installed models. The core advantage of this tool lies in its deep integration with the content-addressable logs introduced in LLM 0.32rc1, enabling efficient handling of multi-turn conversation requests through hash-based deduplication for optimized message storage and transmission. Whether developers are quickly testing models or building local AI applications, this plugin provides a simple, standard API interface, reducing the complexity of integrating with the OpenAI ecosystem.
Key Features
- OpenAI-Compatible API: After startup, it provides a /v1/chat/completions endpoint fully compatible with the OpenAI Chat Completions request format, supporting parameters such as model and messages, allowing seamless switching for existing applications.
- Local Model Exposure: Automatically exposes all models installed via LLM, enabling API calls without additional configuration and supporting unified management of multiple models.
- Content-Addressable Log Optimization: Utilizes the new schema design of LLM 0.32rc1 to hash-deduplicate message parts in multi-turn conversations, reducing redundant data transmission and improving performance in long-conversation scenarios.
- Simple Deployment: Install via uv or pip, start the server with a single command, and support custom ports (e.g., -p 9001), suitable for local development and testing.
Use Cases
- Local development and debugging: Quickly start an OpenAI-compatible API service to test model responses or integrate into development environments.
- Unified interface for multiple models: Expose multiple installed models on one server, switch via the model parameter, making it easy to compare outputs from different models.
- Teaching and demonstrations: Show how the Chat Completions API works to students or teams in offline environments.
- Automated testing: Provide a stable local API endpoint for CI/CD pipelines to simulate OpenAI services for integration testing.
Pros
- Seamless integration with the LLM ecosystem, simple installation, and minimal configuration.
- Fully compatible with the OpenAI API, resulting in low migration costs.
- Leverages content-addressable logs to optimize storage and transmission efficiency in multi-turn conversations.
- Open source and maintained by a well-known developer, with an active community and high reliability.
Pricing
This plugin is completely free, licensed under the open-source MIT license. Users can install it directly via uv or pip, with no subscription or usage fees.
Summary
llm-chat-completions-server is a lightweight, efficient local API server plugin designed for LLM users, providing an OpenAI-compatible Chat Completions endpoint. It optimizes multi-turn conversation processing through content-addressable log technology while simplifying the exposure and invocation of local models. For developers, researchers, or anyone needing a local API service, this tool is a practical and free choice, especially suitable for use with the LLM ecosystem.
Version History
- llm-chat-completions-server 0.1a0 发布 (2026-07-30): Simon Willison released the llm-chat-completions-server 0.1a0 plugin, which can start a server compatible with the OpenAI Chat Completions API on local port 9001, exposing all installed models in the LLM tool.