Run a local AI model on your PC — private, offline, no account
A local AI (or local LLM) is a model that runs entirely on your own computer: you download it once and every answer is computed on your hardware. No account, no logging, no per-token bill, no content filters added by a provider — and it keeps working with the internet unplugged. Strata is the easiest way to get one on Windows: one package, one model picker, done.
Prompts and files never leave the PC — nothing to log, leak or train on
No account
No sign-up, no email, no API key, nothing to ban
No restrictions
Uncensored by design: no provider moderation layer; the only rules are the model's own training
No bills
Free software, free models — no subscription, no per-token pricing
Offline work
Plane, cabin, outage: it runs the same with the network off
Your tools
An OpenAI and Anthropic compatible API on localhost for editors and agents
Set it up in a minute
Download Strata 0.2.17, extract the zip and run the exe. Allow the administrator prompt.
Setup downloads the app and the right engine for your hardware — NVIDIA, AMD, Intel or plain CPU. You never compile anything.
Pick a model: Qwen3.8-Flash-Next is the recommended one. Small quants (Q2_0, IQ2_XS) run on 8–16 GB of RAM; bigger quants answer better.
Chat in the app, script against strata in the Terminal, or point your coding tools at the local API.
What it looks like
A private chat with a local model — no account, no cloud in sight.Watch your local LLM work: tokens per second, memory use, request slots.
FAQ
What is a local AI model?
An LLM that runs entirely on your own computer instead of a provider's servers. You download the model once and every answer is computed on your hardware: no account, no logging, no usage fees, and it works with the internet unplugged.
Is a local AI uncensored?
A local model answers to you alone: no provider-side moderation, no content logging, no account to ban. The only limits come from how the model itself was trained. You pick the model, you set the rules.
Can I run a local LLM without a GPU?
Yes. Small quants of a modern model run fine on an ordinary CPU with 8–16 GB of RAM. An NVIDIA, AMD or Intel graphics card speeds things up a lot, but it is optional.
Is a local AI model really free?
Yes. Strata is free and open source under the MIT license, and the models are published by their providers at no cost. No subscription, no per-token bill, no API key.
Does it work offline?
After the one-time download of the app, engine and model, yes: chat, code and image work all run with no network connection at all.
Can my coding tools use a local LLM?
Yes. Strata exposes an OpenAI and Anthropic compatible API on localhost, so editors, agents and scripts that speak either API use your local model with no key and no bill.
Only trust this page. Fake AI tool builds float around forums and file sharing sites. The real build matches what the maintainers publish on GitHub.