# Use AI on your own device

Editorial revision: 2026-10-28. Product documentation dates and access limits are shown on individual listings.

> Which AI tools can run models on my own device?

Local model tools trade cloud dependence for downloads, device memory and setup. Choose between a desktop interface, a runtime/API and a document-oriented workspace, then verify the path your data takes.

A hand-picked shortlist from the Agent Search Engine index. Every pick is a real listing; the repository, pricing-category and activity figures are recorded catalogue fields, not performance tests. Browse on the web: https://agentsearchengine.app/use/use-ai-on-your-device

## Editorial starting points

### Desktop model workspace: LM Studio
- Free local app; hardware and model terms · Platform · License not checked
- **Setup:** Desktop — model download / hardware
- **Why:** LM Studio supplies a desktop chat interface and local model server. Download a compatible model first and check hardware requirements; the separate Bionic app has a different workflow and cloud options.
- [Capability source](https://lmstudio.ai/docs/app/offline)
- [listing](https://agentsearchengine.app/agents/lm-studio) · [official](https://lmstudio.ai/docs/app)

### Runtime and API: Ollama
- MIT runtime; local hardware/model costs; paid cloud use · Infrastructure · Open source · recorded license: MIT
- **Setup:** Runtime — model download / configuration
- **Why:** Ollama runs supported downloaded models through an app, command line and API. Its optional cloud models have a separate data route, and individual model licenses can restrict use.
- [Capability source](https://docs.ollama.com/faq)
- [listing](https://agentsearchengine.app/agents/ollama) · [official](https://ollama.com/)

### Documents and workspaces: anything-llm
- Local software; hardware/provider costs; paid options · Platform · 68k★ · JavaScript · Open source · recorded license: MIT · repository push 3 days ago
- **Setup:** Desktop / self-host — provider setup
- **Why:** AnythingLLM combines document chat and agent workflows. Desktop and Docker serve different needs; selecting remote models, embeddings or transcription changes processing location and operating costs.
- [Capability source](https://docs.anythingllm.com/installation-desktop/overview)
- [listing](https://agentsearchengine.app/agents/anything-llm) · [official](https://anythingllm.com/)

## What to watch for
Check the selected model's license, memory needs and supported capabilities. Complete the necessary downloads before testing offline. Inspect embeddings, search, transcription, telemetry and connectors separately: local chat does not establish that every surrounding feature stays on your device.

---
Source: [Agent Search Engine](https://agentsearchengine.app) — more task-first guides: https://agentsearchengine.app/use