Sider
AssistantsRead pages with Chat, handle browser tasks with Hand and customize your browsing experience with Code. Sider adds an AI workspace to the sites and documents you already use.
Chat beside the sourceHand browser tasks
LOCAL AI / DESKTOP & PROJECTS
Jan
A free desktop AI assistant that runs supported models locally, organizes project files and can expose a local API for your own applications.
Official Jan Desktop getting-started screenshot with chats, projects and the model selectorExplore the image ↓
What you can do with it
What this tool is for
Jan makes local language models approachable through a desktop chat application. Instead of starting with a command-line server, you can download a supported model, open a conversation and gradually add files or project instructions. Local chat does not require a service account, and inference on your own machine avoids per-request cloud API charges.
The practical tradeoff is capacity. A model must fit the available memory, and a downloaded model is not automatically as capable as the largest hosted service. Jan is a good starting point for personal notes, rewriting, bounded document questions and developers experimenting with a local API. For very long documents or demanding reasoning, evaluate a representative task before treating a small model as a complete replacement.
01
Private first drafts
02
One-project document assistant
03
Offline travel preparation
04
Prototype a local integration
A closer look at the product
Official product image. Click to inspect the details.
Product interface and examples
Key features, practical tasks
Capabilities in practice
- Model selection with hardware context: The model hub helps indicate which choices fit, may be slow or will not fit the device. Start with a modest model and increase size only when better answers justify the additional memory and latency.
- Supported local formats: Current documentation covers GGUF models through llama.cpp and MLX options for Apple Silicon. Use a model format and runtime compatible with the installed version rather than assuming every downloaded checkpoint will load.
- Project-level organization: Projects keep related conversations, files and instructions together. A shared set of background materials can support several questions without rebuilding the project context in every new chat.
- Document and media inputs: Upload supported documents, images or audio according to the selected model. A text model cannot gain vision or audio understanding simply because the interface accepts an attachment.
- Optional cloud connections: Configure a supported cloud provider when a task needs a different capability. That changes the data path and can introduce provider charges even though the Jan application itself remains free.
- Local API server: Expose a local OpenAI-compatible endpoint for development. The documented default address is 127.0.0.1:1337/v1; configure the required API authentication and keep local experiments bound to the intended interface.
- Tool integration: MCP connections can add capabilities to supported workflows. Choose permissions for a concrete task and review tool behavior before allowing it to write files or contact external systems.
Get started, step by step
- 01
Install the correct desktop build
- 02
Download a model that fits
- 03
Create a focused project
- 04
Check output and data routing
A practical first workflow
A practical first workflow
- Install the correct desktop build. Use an official release for the operating system and processor. Allow space for the application and the separate model downloads.
- Download a model that fits. Use the hub guidance and start small. Let the download complete, then verify a short ordinary conversation before adding files.
- Create a focused project. Add instructions and one short reference file. Specify what the assistant should do when information is absent rather than asking it to fill every gap.
- Check output and data routing. Review facts in the result and confirm whether the selected model and tools are local. Save the final document separately from exploratory conversations.
Example prompt or task: Rewrite these notes into a clear project brief. Preserve names, dates and numbers exactly. Separate confirmed requirements from open questions, and do not add facts that are absent from the supplied notes.
This is a suggested evaluation task, not a performance claim. Check preservation of important details before increasing document length or selecting a larger model.
Find the right plan for you
Free desktop app
Explore official plans ↗Free software Local models use your hardware
The app is free. Model licenses, device requirements and optional cloud charges are separate.
What is free and what can cost extra
The desktop application is available under Apache-2.0. Running a local model does not require paying a cloud provider for every response, but it uses storage, RAM or graphics memory, electricity and time on your own machine. Individual downloaded models can have different licenses.
Cloud model integrations remain optional and may require a separately billed API key. Network tools and search can also depend on outside services. The free classification here is based on the local desktop route, not on an assumption that every connected provider is free.
A few things worth knowing
Frequently asked questions
Do I need an account for local chat?
The local desktop route does not require an account. Optional remote providers and services can have their own sign-in or API requirements.
Can Jan work offline?
Local inference can work after the required application and model downloads. Cloud providers, web search and remote tools still need a network connection.
Will any model from the internet work?
No. The format, runtime, model architecture and device capacity must be compatible. Use the documented model hub or supported import workflow.
Can I ask about PDFs?
The application supports document workflows, but extraction and retrieval quality still matter. Check whether the relevant text was processed and whether the answer is supported by it.
Can a text-only model analyze a photo?
Not merely because the photo can be attached. Choose a model that supports the relevant vision input, and verify its actual interpretation.
Is its API a free hosted API?
No. It is a server running on your machine. You supply compute and must configure access; binding it to other interfaces changes who can reach it.
Is every conversation guaranteed to remain local?
Only when the selected model and all relevant processing and tools are local. A configured cloud provider can receive the request and associated content.
The details behind your decision
Explore platforms, inputs and outputs, licensing, and access requirements.
Four useful starting projects
Four useful starting projects
- Private first drafts: Rewrite notes with a local model and no remote tools. Compare the result with the original so names, amounts and important uncertainty are not silently changed.
- One-project document assistant: Create a project around a small collection of related files. Ask for a summary and unresolved questions, and check each claim against the actual file.
- Offline travel preparation: Download a suitable model before losing connectivity. Use it for reorganizing saved notes, not for pretending to know current transport changes or live availability.
- Prototype a local integration: Point a small development client at the local API. Test a short request and an error case before replacing any production provider configuration.
Platforms, inputs, outputs and setup
Platforms, inputs, outputs and setup
Desktop releases support Windows, macOS and Linux. Hardware needs depend on model size, quantization, context and runtime; there is no universal memory number that guarantees every model will work well. The hub guidance is a starting point, followed by a short task on your actual device.
The file guide covers common document formats including PDF, Markdown and Office files, along with code and supported media. Project documents are processed for retrieval. Image and audio understanding require suitable models, and a large attachment does not mean the entire file necessarily appears in every answer context.
Data, permissions and licensing
Data, permissions and licensing
For local models, the desktop documentation describes local processing. Selecting a cloud model or remote tool changes that boundary, including content needed to handle a file request. Check the full workflow before placing sensitive material in a project.
Use narrowly scoped tool permissions. Do not expose a development API to a network without the intended authentication and access controls. Apache-2.0 applies to the application code; the terms of a downloaded model and the rights to your files remain separate.
Sources, documentation and help
Official sources and review date
Official sources and review date
Reviewed October 3, 2026. Product facts come from the official sources below. Suggested projects, prompts and review methods are FindGoodAI editorial guidance, not measured performance results.
Your experience can help someone choose
COMMUNITY
Sign in to rate and comment
Tried it? Share your experience
Loading experiences…
Manage my comments
Community and moderation
Saved tools are private. Approved comments are public; edits return to moderation. Comment counts include approved comments and replies. Each person has one rating. New ratings are automatically approved; withdrawn, unapproved or invalid ratings do not count. Share real experiences and avoid spam, private information or personal attacks.
For moderation appeals or data requests: support@findgoodai.com
NEXT STEP
Explore Jan ↗Try a small local model on a real task
Bring a real task and see how it fits the way you work.
Jump to a section
User experiences