Privacy by design
No accounts, no tracking, no server-side logs. Conversations live only in your browser and never touch a database.
Open-source · Privacy-first · LiteLLM-powered
AI that answers you,
not about you.
OpenThena is a fully open-source AI assistant powered by a LiteLLM-style gateway and the Xiaomi MiMo brain. It streams answers in real time — and keeps every conversation on your device. No accounts, no tracking, no stored chats.
Zero-storage architecture · your chats never touch a database
www.openthena.com/chat
Is my conversation stored anywhere?
Nope. Your messages stream straight to the model and back — nothing is written to any server or database. This chat lives only in your browser. ✨
Why OpenThena
Powerful AI, private by default
Everything you expect from a modern assistant — without the surveillance, lock-in, or hidden telemetry.
Fully open source
Every line is MIT-licensed and auditable. Fork it, self-host it, and verify exactly what runs on your data.
LiteLLM-style gateway
A unified, OpenAI-format gateway routes to any model. Swap providers with a single environment variable.
Xiaomi MiMo brain
Reasoning powered by MiMo v2.5 — fast, capable, and multilingual — with room to plug in any model you trust.
Real-time streaming
Answers stream token-by-token over the edge, so you read the response as it is being written.
Deploy anywhere
One-click to Vercel or self-host the whole stack. No vendor lock-in, no hidden telemetry.
How it works
Three steps, zero footprints
01
Ask anything
Type your prompt. Nothing is written to a server log — your message goes straight to the model gateway.
02
The gateway routes it
OpenThena's LiteLLM-style gateway forwards your request, in OpenAI format, to the Xiaomi MiMo brain on the edge.
03
Stream & forget
The answer streams back instantly and is stored only in your browser. Close the tab to wipe the slate clean.
The privacy promise
We store nothing.
Most AI products log every prompt to improve their models — and their bottom line. OpenThena takes the opposite stance: the server is a stateless relay. Your words pass through and vanish.
●logs_written = 0
No database
Conversations are never persisted server-side. There is no table to leak.
No tracking
No analytics, no fingerprinting, no third-party scripts watching you.
No cookies
We don't set advertising or identity cookies. Nothing follows you around.
Local-only history
Your chats live in your browser's storage and clear when you say so.
Under the hood
A stack you can actually read
No black boxes. Every layer is open source and documented.
Unified model interface
LiteLLM-style gateway
One OpenAI-format gateway in front of every model. Swap providers with a single env var — your UI never changes.
The reasoning brain
Xiaomi MiMo
MiMo v2.5 delivers fast, capable, multilingual reasoning through an OpenAI-compatible endpoint.
Edge delivery
Next.js on Vercel
Server-side streaming on the edge keeps responses instant and the deployment footprint tiny.
Open source, end to end
Audit it, fork it, improve it. OpenThena is MIT-licensed and built entirely in the open — because trust requires transparency.
MIT License Star on GitHub Fork & self-host
FAQ
Questions, answered
Yes. There is no database and no authentication. Your conversations are kept in your browser's local storage only. The API route is stateless and writes no logs — it simply forwards your request to the model and streams the answer back.
Ask freely. Stay private.
Open the app and start a conversation. No sign-up, no setup — just you and an AI that forgets on purpose.