LogoAIAny
Icon for item

Jan

Jan is an open-source ChatGPT replacement that runs locally on your computer. It lets users download and run LLMs (Llama, Gemma, Qwen, GPT-oss, etc.) from HuggingFace, connect to cloud models (OpenAI, Anthropic, Mistral, etc.), create custom assistants, and expose an OpenAI-compatible local API for integrations. Privacy-first design and support for MCP make it suitable for local and hybrid workflows.

Visit Website
Visit Website

Introduction

Jan — Open-source ChatGPT replacement

Overview

Jan is a desktop-first, open-source application that provides a privacy-focused way to run and interact with large language models (LLMs). It supports both fully-local model execution and optional cloud integrations, offers custom assistant creation, and exposes an OpenAI-compatible local server for other apps to use.

Key features
  • Local LLMs: download and run models (Llama, Gemma, Qwen, GPT-oss and others) from HuggingFace and compatible repositories.
  • Cloud integrations: connect to OpenAI, Anthropic (Claude), Mistral, Groq, and similar providers when cloud inference is desired.
  • Custom assistants: build specialized assistants with custom prompts, tools, and behaviors.
  • OpenAI-compatible API: runs a local API (default at localhost:1337) so other apps can call Jan as if it were an OpenAI endpoint.
  • MCP support: integrates Model Context Protocol (MCP) to enable agentic capabilities and richer tooling.
  • Cross-platform desktop apps: builds available for Windows, macOS, and Linux (AppImage / .deb). Official distribution via jan.ai, Microsoft Store, and Flathub.
  • Privacy-focused: everything can run fully offline; network calls only occur for chosen cloud providers or optional updates.
Installation & quick start

Jan provides prebuilt desktop packages and an easy developer workflow:

  • Download installers from jan.ai or GitHub Releases (Windows, macOS, Linux AppImage/.deb).
  • For development: Node.js ≥ 20, Yarn, Make and Rust (for Tauri) — clone repo and run make dev.
  • The app handles model downloads and local runtime management; GPU acceleration supported when available.
Architecture & models

Jan is a desktop application using a native wrapper (Tauri) with a local core to manage model runtimes and extensions. It can run inference on CPU or GPU (depending on your hardware and model), and supports popular community runtimes (e.g., llama.cpp backends). The project exposes extension/plugin points and a local HTTP API compatible with OpenAI’s HTTP semantics for broad interoperability.

Privacy, offline use, and performance

Jan emphasizes privacy-first operation: users can choose to run models 100% offline so prompts and data never leave their machine. For larger models or specialized capabilities, Jan also supports hybrid workflows where cloud providers are used selectively. System requirements are documented (e.g., RAM guidance for 3B/7B/13B models) and GPU acceleration is supported for better performance.

Integrations & extensibility
  • OpenAI-compatible local API makes it easy to plug Jan into existing tools and scripts.
  • MCP and plugin support enable agentic workflows and custom tool integrations.
  • Community extensions and config files let users add new model sources, assistants, or UI behaviors.
Use cases
  • Local/private chat assistants for sensitive data.
  • Offline prototyping with open LLMs.
  • Hybrid setups that route heavy tasks to cloud models while keeping sensitive work local.
  • Developer toolchains that need an OpenAI-compatible endpoint without using external keys.
Community, contributing & support

Jan is actively developed on GitHub (janhq/jan) and maintains documentation at jan.ai/docs. Community support is available via Discord and GitHub issues. The project is open to contributions — build instructions, testing, and contribution guidelines are included in the repo.

License

Apache 2.0 — permissive open-source license that allows use, modification, and distribution.

Where to learn more
Back

Information

  • Websitegithub.com
  • Authorsjanhq (GitHub organization), Jan team, Community contributors
  • Published date2023/08/17

Categories

Tags

More Items

Icon for item

OWL: Optimized Workforce Learning for General Multi-Agent Assistance in Real-World Task Automation

2025
CAMEL-AI

OWL is an open-source multi-agent collaboration framework from CAMEL-AI for building an AI 'workforce' to automate complex real-world tasks. It includes rich toolkits (browser automation, document parsing, multimodal analysis, code execution), MCP (Model Context Protocol) integration, local web UI, and examples. OWL focuses on parallel execution, privacy-first design, and customizable multi-agent workflows, and achieved top open-source performance on the GAIA benchmark.

ai-agentai-frameworkmcpai-workflowgithub+2
Icon for item

MLX LM

2025
ml-explore (GitHub organization)

MLX LM is a Python package to run, generate with, and fine-tune large language models on Apple Silicon using MLX. It integrates with the Hugging Face Hub, supports quantization and uploading of models, low-rank and full-model fine-tuning (including for quantized models), distributed inference and training, streaming generation, sampling/custom logits processors, prompt caching, and a convenient CLI and Python API.

llmhuggingfacegithubai-libraryai-inference+4
Icon for item

opcode

2025
Asterisk / getAsterisk (project maintainers)

opcode is a desktop GUI application and toolkit for Claude Code. It provides visual project and session management, custom Claude Code agents, secure background agent execution, usage and cost analytics, MCP server management, session timelines with checkpoints, and a built-in CLAUDE.md editor. Built with Tauri (Rust) and React, it emphasizes local-first storage, process isolation, and per-agent permissions.

claudeanthropicmcpmcp-clientai-client+3