Cherry Studio
A free, open-source multi-model AI desktop client — chat, agents, and knowledge bases in one window.
About
Cherry Studio is the rare desktop AI client that grew into a genuine agent workbench. One window hosts every major model provider — OpenAI, Anthropic, Google, Azure, plus local models via Ollama, LM Studio, or LocalAI — with 300+ pre-configured assistants, a personal knowledge base, AI drawing, translation, and side-by-side multi-model answers. Version 2.1.5 (October 2026) added phone-to-PC file uploads, MCP SDK v2, and a System Doctor that proposes reversible config fixes.
It is fully open source (AGPL-3.0), ships on Windows, macOS, Linux, and iOS, and stores your data locally — which is why it has become the default recommendation for people who want a private, no-subscription AI workspace. With ~52,700 GitHub stars and same-day release cadence, it is one of the most actively maintained clients in the space.
The honest caveat: Cherry Studio has no models of its own — you bring every API key and pay every provider bill yourself, so it is a shell, not a service. English localization still has rough edges in places. But as a free, private, multi-provider front end, nothing else is close on features per dollar (zero dollars).
What it can do
- Unified multi-provider chat: OpenAI, Anthropic, Gemini, Azure, Ollama, LM Studio, LocalAI
- 300+ pre-built assistants plus custom assistant creation
- Agent features: MCP tools, autonomous coding, Pi agent runtime with sandboxed code mode
- Knowledge bases, AI drawing, translation, Mermaid charts, WebDAV backup
- Local data storage for privacy; phone-to-PC remote access
- AGPL-3.0 licensed; cross-platform (Windows, macOS, Linux, iOS)
Setup
- Download the installer for your OS from cherry-ai.com (Windows, macOS, Linux, iOS).
- Open Settings → Model Services and add a provider with your API key (or point at a local Ollama/LM Studio endpoint).
- Pick a model, choose one of the 300+ built-in assistants or create your own.
- Optionally connect MCP servers and build a personal knowledge base for retrieval.
Prerequisites
A desktop or mobile OS; at least one LLM provider API key (or a local model runner). No account or subscription needed.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Cherry Studio, 52,668 GitHub stars; widely adopted desktop AI client.; Trust data reconstructed by coordinator after parallel-worker file overwrite; scores evidence-based from entry metadata..
★ 52,668 GitHub stars
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.