Two open models.
One flat price.
Talk to Qwen 3.6 and DeepSeek V4 — frontier open-source AI, served from Switzerland. Switch between them mid-conversation. Sleep well.
Private by design
Your conversations are never stored on our servers — they live on your device, and only there. We couldn't read them if we wanted to.
Hosted in Switzerland
Inference runs on Swiss servers with no prompt logging. Your words are processed, answered, and forgotten.
Open-source models
Qwen 3.6 and DeepSeek V4 — open weights anyone can inspect. No black boxes, and your chats train no one's next model.
- Both models, unlimited conversations
- Streaming answers, model switch mid-chat
- Images, PDF & file uploads
- Web + CLI access
- Cancel anytime
The most private way to use AI
We don't have a user database — your subscription lives at Stripe, and logging in just checks it's active. We don't store conversations — they stay in your browser, on your device. We don't run trackers, analytics, or ad pixels. Inference is Swiss-hosted with no prompt logging. There is simply nothing here to leak, sell, or subpoena.