omnifeed Search

Jan

Open-source desktop app for running AI models locally or connecting to cloud models

Visit jan.ai

In our rankings

Scores on the apps for running ai models locally criteria

Evaluated Nov 5, 2026
  • Model formats and downloads ×3 5

    5/5 means: Two or more model formats, in-app Hugging Face search and download, and import of local model files

    Runs GGUF models and MLX models on Apple Silicon, downloads through a hub connected to Hugging Face, and imports local files in both formats.

    4 quotes
  • GPU support ×3 5

    5/5 means: NVIDIA, AMD, Apple Silicon and Intel GPUs (for example through Vulkan), and a documented processor-only mode

    GPU support for NVIDIA, AMD and Intel Arc on Windows, engine builds for Apple Metal and Vulkan, and a CPU engine.

    2 quotes
  • Local API for other apps ×2 3

    3/5 means: OpenAI-compatible chat endpoint that only listens on your own computer

    An OpenAI-compatible chat endpoint that can be opened to the local network; embeddings endpoints are not documented on the pages we captured, so it does not meet the bar for 4.

    3 quotes
  • Release recency ×2 5

    5/5 means: Latest release less than 1 month old

    The newest changelog entry, Jan v0.8.8, is dated 28 October 2026.

    1 quote

Evidence (17 quotes)

  1. 1
    “these models are in a format called GGUF.”
    docs.jan.ai/desktop/manage-models · checked
  2. 2
    “Import MLX-Swift models for Apple Silicon.”
    docs.jan.ai/desktop/manage-models · checked
  3. 3
    “The easiest way to get started is using Jan's built-in model hub (which is connected to HuggingFace's Model Hub”
    docs.jan.ai/desktop/manage-models · checked
  4. 4
    “Import GGUF model files for local inference on CPU or GPU.”
    docs.jan.ai/desktop/manage-models · checked
  5. 5
    “Windows: 10+ with GPU support for NVIDIA/AMD/Intel Arc”
    github.com/janhq/jan · checked
  6. 6
    “Engine variants are picked with JAN_ENGINE_VARIANT (tokens: cpu, vulkan, metal, cuda12, cuda13, hip/rocm, joined by -)”
    github.com/janhq/jan · checked
  7. 7
    “macOS: 13.6+ (8GB RAM for 3B models, 16GB for 7B, 32GB for 13B)”
    github.com/janhq/jan · checked
  8. 8
    “Jan provides a built-in, OpenAI-compatible API server that runs entirely on your computer”
    docs.jan.ai/desktop/api-server · checked
  9. 9
    “The server is accessible from other devices on your local network (e.g., your phone or another computer).”
    docs.jan.ai/desktop/api-server · checked
  10. 10
    “The chat completions endpoint would be http://127.0.0.1:1337/v1/chat/completions”
    docs.jan.ai/desktop/api-server · checked
  11. 11
    “You'll choose tracking preferences at first launch. Change them anytime in Settings.”
    docs.jan.ai/desktop/privacy · checked
  12. 12
    “Jan will never peek at your chats, settings, or model choices.”
    docs.jan.ai/desktop/privacy · checked
  13. 13
    “Open source : Apache 2.0 licensed, built in public.”
    docs.jan.ai/desktop · checked
  14. 14
    “macOS, Windows, and Linux apps with full offline support.”
    docs.jan.ai/desktop · checked
  15. 15
    “Cloud Integration: Connect to GPT models via OpenAI, Claude models via Anthropic, Mistral, Groq, MiniMax, and others”
    github.com/janhq/jan · checked
  16. 16
    “October 28th, 2026 Jan v0.8.8: Security Fixes, llama.cpp 0.7.0 & Local API Trusted Hosts”
    jan.ai/changelog · checked
  17. 17
    “Zero data collection until you say so.”
    docs.jan.ai/desktop/privacy · checked

Alternatives