new-api MCP Server: LLM Gateway for OpenAI, Claude, and More
new-api: Unifying LLM Access for Internal Distribution
The new-api MCP Server, a JavaScript-based project with over 45,000 GitHub stars, functions as an AI model interface management and distribution system. It's designed to streamline access to various large language models (LLMs) by converting them into a unified calling format, supporting popular interfaces like OpenAI and Claude. This makes it a practical solution for individuals or enterprises managing and distributing AI model access internally.
A Unified LLM Gateway
At its core, new-api acts as a next-generation LLM gateway. It takes the complexity out of integrating with different AI models by providing a consistent API layer. This means developers can write against a single interface, and new-api handles the translation to the underlying model's specific requirements. This capability is particularly useful for environments that consume multiple LLMs and need to standardize their interaction.
The project is built as a secondary development of One API, indicating a focus on extending and refining existing solutions for LLM management. It's available as a Docker image on both ghcr.io and DockerHub, simplifying deployment for containerized environments.
Internal AI Asset Management
Beyond just unifying model access, new-api positions itself as an AI asset management system. For organizations, this translates to better control over how internal teams access and utilize AI resources. Instead of each team or application needing to manage separate API keys and integration logic for every LLM, new-api centralizes this. This approach helps in managing costs, enforcing access policies, and providing a consistent developer experience across an organization's AI initiatives.
The system's support for multiple languages in its overview suggests a global usability intent, though the primary development language is JavaScript.