#3030th of 49 in Desktop AI apps

local-studio

Desktop coding-agent UI, forked from T3 Code, with a local model gateway

Stars
1.8k
License
MIT
Last commit
Oct 2026
Last release
Oct 2026
Language
TypeScript
Stars, last 29 weeks+1,405
21 Apr 2026: 440 stars28 Apr 2026: 483 stars5 May 2026: 818 stars12 May 2026: 908 stars19 May 2026: 953 stars26 May 2026: 1,011 stars2 Jun 2026: 1,060 stars9 Jun 2026: 1,090 stars16 Jun 2026: 1,126 stars23 Jun 2026: 1,155 stars30 Jun 2026: 1,260 stars7 Jul 2026: 1,305 stars14 Jul 2026: 1,334 stars21 Jul 2026: 1,459 stars28 Jul 2026: 1,493 stars4 Aug 2026: 1,542 stars11 Aug 2026: 1,624 stars18 Aug 2026: 1,700 stars25 Aug 2026: 1,726 stars1 Sept 2026: 1,734 stars8 Sept 2026: 1,743 stars15 Sept 2026: 1,753 stars22 Sept 2026: 1,764 stars29 Sept 2026: 1,807 stars6 Oct 2026: 1,817 stars13 Oct 2026: 1,827 stars20 Oct 2026: 1,832 stars27 Oct 2026: 1,835 stars3 Nov 2026: 1,841 stars10 Nov 2026: 1,845 stars

440 on 21 Apr 20261,845 on 10 Nov 2026

From GitHub's daily star counts.

Checked on GitHub.

Overview

Local Studio is a thin fork of T3 Code, an Electron app that runs coding agents such as Codex, Claude, Cursor, OpenCode, Pi and Oh My Pi. It adds a bundled controller that discovers vLLM, SGLang, llama.cpp and other /v1/models servers and exposes them through one OpenAI/Anthropic-compatible gateway. Settings can also link machines on a Tailscale tailnet into a fleet.

Who it is for: Self-hosters running local models who want a coding-agent desktop app

Strengths

  • Controller serves /v1/chat/completions, /v1/messages and /v1/responses across machines
  • Installs a controller on Linux or macOS tailnet machines over ssh as a user service
  • Can launch pinned registry recipes on free NVIDIA GPUs
  • Local changes capped at 4,000 lines by a check script, easing upstream merges

Weaknesses

  • Needs at least one agent CLI (pi, omp, or another provider) installed and logged in first
  • Windows installer is not code-signed, so SmartScreen warns
  • Fleet install requires key-based ssh and Tailscale; Linux and macOS targets only
  • Most docs are upstream's and may not describe Local Studio additions

What it needs

  • GPU optional
  • Needs Tailscale (for fleet), ssh (for remote controller install), Agent CLI such as pi or omp, Node 24, pnpm 11, Bun 1.3.14 (development only)
  • Models: vLLM, SGLang, llama.cpp, OpenAI-compatible /v1/models servers
  • port 18091

Quick answers

Is local-studio open source?

Yes. Its code is on GitHub under the MIT license.

Does local-studio need a GPU?

No. It runs on a CPU and can use a GPU when you have one.

Also in Desktop AI apps

See all 49
Also in Desktop AI apps
RankProjectScore
1cherry-studioDesktop client for chatting with many LLM providers, with agents and knowledge bases53.4k stars+1,124 (+2.1%) in 30 daysAGPL-3.092 out of 100
2dsh-desktopNative Windows and macOS desktop client for DeepSeek Harness, built on plugins35.7k stars+5,734 (+19%) in 30 daysMIT90 out of 100
3screenpipeRecords screen and audio locally, makes it searchable for you and your agents22.3k stars+481 (+2.2%) in 30 dayscustom license90 out of 100
#4prompt-optimizerPrompt rewriting and testing tool for web, desktop, Chrome and MCP Live demo ↗ (opens in a new tab)41.2k stars+3,703 (+9.9%) in 30 dayscustom license86 out of 100
#5aionuiDesktop and WebUI app that runs a built-in agent plus external CLI agents34.2k stars+923 (+2.8%) in 30 daysApache-2.086 out of 100