FreeListed
LM Studio
LM Studio
A polished desktop app for running open AI models on your own computer — point-and-click instead of a terminal, free for personal and work use, private and fully offline.
Tell it the task. Get the tool.
Describe what you want to do in your own words. The finder matches your task against 139 hand-checked tools and explains why each one fits — no invented capabilities, only what the directory has verified.
Best for: skipping category browsing entirely — start from the job, not the tool type.
Try:I need a free AI coding assistantWhich AI can analyze large PDF documents?I want to create music with vocals for freeFree AI research tool with citations for my thesisI want to run AI models locally for privacyMake presentations from an outline without paying
4 matches for “I want to run AI models locally for privacy”
ListedWhy it matches: runs locally / privacy-first · open weights you can run yourself.Full LM Studio listing →
FreeListed
Jan
Menlo Research
A free, open-source desktop app that runs AI models fully offline on your own computer — a private, self-hosted ChatGPT alternative that can also connect to cloud models, from jan.ai.
ListedWhy it matches: runs locally / privacy-first · open weights you can run yourself.Full Jan listing →
ListedWhy it matches: runs locally / privacy-first · open weights you can run yourself.Full Ollama listing →
ListedWhy it matches: runs locally / privacy-first · open weights you can run yourself.Full Whisper listing →
How the finder decides
The finder is deliberately simple. Your description is matched against a fixed set of task signals — writing, coding, research, images, video, music, transcription, presentations, meetings, privacy-sensitive work, local execution, and budget constraints. Each signal maps to the verified listings that genuinely serve that task, and every result shows the exact signals that fired, so a recommendation is never a black box.
It is a rule-based matcher over the hand-checked database, not a language model — that is the point. It cannot invent tools, hallucinate capabilities, or recommend anything that is not in the verified catalogue. When nothing in the directory confidently fits, it says so instead of guessing. The trade-off is honest: describe the task the way you would to a colleague, mentioning what you make, the format you need, and whether it must be free.
Matching detail: your words are matched against task signals mapped to verified listings; every recommendation cites the signals that triggered it. It is a transparent rule-based matcher over the hand-checked database — not a language model, so it cannot invent tools or capabilities that are not listed.