An open-source multi-provider (including local) fork of Gemini-CLI (github.com)

LLxprt Code is an open-source, multi-provider fork of Google’s Gemini CLI that significantly expands its flexibility and usability for AI/ML practitioners. Building on Gemini’s foundation, LLxprt Code integrates direct support for multiple large language model (LLM) providers including OpenAI, Anthropic (Claude), Google Gemini, OpenRouter, Fireworks, and local models like LM Studio and llama.cpp. This multi-provider approach allows users to seamlessly switch between models and services during a session, enabling diverse workflows from code querying to multimodal app generation with PDFs or sketches, all within a consistent, themable CLI environment. Technically, LLxprt Code preserves full Gemini CLI compatibility, including Google authentication, while offering enhanced configuration options such as advanced model parameter tuning, ephemeral settings, and profile management. Notably, it supports local model hosting for privacy-sensitive tasks, allowing connection to any OpenAI-compatible server and facilitating high context limits for large-scale processing (e.g., 100,000 token context windows). The tool also enables integration with MCP servers to connect additional capabilities like media generation and real-time grounding via Google Search. Its sophisticated prompt customization system further empowers users to control AI behavior across providers and use cases. LLxprt Code thus represents a powerful, open platform for flexible, multi-model experimentation and production AI workflows, useful for developers seeking privacy, provider diversity, and deep customization in LLM-driven projects.