LLxprt Code
An open-source, multi-provider fork of Gemini CLI — use whatever LLM you want to code in your terminal.
About
LLxprt Code is an open-source fork of Google's Gemini CLI rebuilt around provider freedom: Anthropic, Gemini, OpenAI-compatible endpoints, free tiers (Gemini, Qwen), Claude Pro/Max subscriptions, and fully local models via Ollama, LM Studio, or llama.cpp. No telemetry by default, subagents that can each run different models or providers, MCP integration, a profile system for saved configurations, and a themed terminal UI with a Zed editor integration.
The fork exists for a reason Gemini CLI itself does not serve: people who want the Gemini CLI interaction model without being tied to Google's models and quotas. With commits within days of this writing and Apache-2.0 licensing, it is alive — though the ecosystem is fragmented across several forks, so pick the vybestack upstream (the npm publisher) and pin your install.
The honest caveat: it inherits both the strengths and the rough edges of the Gemini CLI codebase, and 'works with every provider' in practice means 'works best with the providers the maintainers test'. Still, for a free, local-first-capable terminal agent with no account required, it is one of the more practical options.
What it can do
- Multi-provider coding agent: Anthropic, Gemini, OpenAI-compatible, local models
- Free-tier support (Gemini, Qwen) and Claude Pro/Max subscription auth
- Local-first operation with Ollama, LM Studio, llama.cpp; no telemetry by default
- Subagents with per-agent models, providers, and settings
- MCP integration; profile system for saved configurations
- Apache-2.0 licensed
Setup
- Install:
npm install -g @vybestack/llxprt-code(or try withnpx @vybestack/llxprt-code). Node.js 20+ required. - Run
llxprtand choose a provider with/provider(Gemini, Qwen, Anthropic, OpenAI-compatible, or local). - Authenticate per provider (
/auth), or point at a local Ollama/LM Studio endpoint. - Save repeat setups as profiles; create subagents with different models per task.
Prerequisites
Node.js 20+; optionally: a provider API key/subscription, or a local model runner (Ollama, LM Studio, llama.cpp).
Use cases
Works with
Trust score — 8/10
Scored 8/10 (metrics tier, metrics-v1). Evidence: Adoption 6: starScore from 703 stars on vybestack/llxprt-code (log scale vs 298k peer max); Maintenance 9: pushed 2026-10-15 (recency 10); active fork development; Openness 10: Apache-2.0 (0.7*10) + official docs verified (0.3*10); Manual 7: real multi-provider Gemini-CLI fork corroborated by README and npm; docked: fragmented fork ecosystem, provider breadth is partly provider-claimed. License: Apache-2.0.
★ 703 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 6.0/10
- Maintenance · recency, releases, issue load
- 9.0/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · Adoption 6: starScore from 703 stars on vybestack/llxprt-code (log scale vs 298k peer max)
- · Maintenance 9: pushed 2026-10-15 (recency 10); active fork development
- · Openness 10: Apache-2.0 (0.7*10) + official docs verified (0.3*10)
- · Manual 7: real multi-provider Gemini-CLI fork corroborated by README and npm; docked: fragmented fork ecosystem, provider breadth is partly provider-claimed
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.