3,365 apps tracked 0 new today 17 updated today Synced 11m ago

Apps / Jan

Jan

Private offline AI assistant

Menlo Research Pte. Ltd. ✓ Verified

ai.jan.Jan
View on Flathub ↗

Jan is an open source alternative to ChatGPT that runs 100% offline on your computer.

Features

  • Local AI models: Run LLMs such as Llama, Gemma, Qwen and others on your device
  • Optional cloud access: Connect to services like OpenAI, Anthropic, Gemini and Groq when internet-based models are required
  • Custom assistants: Create tailored assistants for writing, coding, summarizing and more
  • OpenAI-compatible API: Includes a local API server at localhost:1337 for tools expecting the OpenAI format
  • MCP support: Integrates the Model Context Protocol for enhanced workflows and tooling
  • Privacy-aware design: Jan works offline by default. Cloud access is optional and user-controlled

Jan is actively maintained by Menlo Research. Feedback, bug reports and feature requests are welcomed through GitHub.

Report an issue ↗Website ↗Source code ↗Subscribe via RSS

Screenshots

Blank Jan page with 0 threads
Blank Jan page with 0 threads
Jan using a search MCP to gather information
Jan using a search MCP to gather information
Assistant creation screen
Assistant creation screen
Jan Hub
Jan Hub
Llama.cpp screen showcasing the models
Llama.cpp screen showcasing the models
MCP Settings
MCP Settings

Observed history

App ChangeVersionWhen ↓
updated0.8.615 Oct 2026

Upstream releases

Release notes supplied by the developer. These dates are not Flathub package publication times.

0.8.69 Oct 2026

A hotfix for 0.8.5: NVIDIA Turing GPUs (GTX 16xx, RTX 20xx) run the CUDA engine again on older drivers, Qwen3.8 with MTP is fast again when Parallel Sequences is auto, and images returned by MCP tools reach the model as images

0.8.58 Oct 2026

A fixes release: the llama.cpp engine (0.5.0) now ships inside the app and runs in its own worker process with CUDA and Vulkan, the jan command-line agent is no longer bundled, the Local API Server enforces Trusted Hosts when bound to 0.0.0.0, and a batch of llama.cpp, MCP, platform and agent fixes

0.8.423 Jul 2026

Backend-managed settings and credentials with OS keyring support, native web_search/web_fetch via tauri-plugin-websearch, built-in extension bundling, per-model chat template kwargs, OpenAI-compatible translating gateway with Responses API, chain-of-thought duration persistence, MCP duplicate serve guard, multi-provider API key rotation, and a large batch of chat, RAG, and desktop fixes

0.8.324 Jun 2026

A visible assistant switcher with a Cmd/Ctrl+J cycle shortcut, a unified reasoning and tool-call trace timeline, per-model sampling defaults for the local API server, a larger 100MB attachment limit, plus a batch of chat, citation, markdown/LaTeX rendering, factory-reset, and Linux fixes