PrivatAI relaunch: faster and with stronger models
PrivatAI has been rebuilt. If you've known us from the early days, you'll remember: we started with a deliberately simple setup — one model, one server, one promise. The promise stayed. Almost everything else got better.
What's new
1. Considerably stronger models
Instead of a single small model, current open frontier models are now available:
- gpt-oss-120b — included in every plan
- glm-5.2 — on the Professional plan
Both speak the OpenAI-compatible API (/api/v1), including streaming, function calling and reasoning. Existing integrations only need to change the model name — nothing else.
2. Faster responses
Thanks to new infrastructure, the models now respond considerably faster — large models without a local GPU, and still without your data ever leaving Europe.
3. The foundation stays
No content logs, no training on your data, GDPR-compliant, processing in the EU. This principle doesn't change — it's the reason PrivatAI exists. New models and faster infrastructure change nothing about how sparingly we handle your data.
4. Ready for your tools
PrivatAI works as a drop-in backend for anything that speaks the OpenAI API — from your own code to OpenClaw and Odysseus. Create an API key, set the base URL, done.
What stays the same
The principle. Your inputs are processed, answered, discarded. No content logs. No training on your data. Processing in the EU, outside the reach of the US CLOUD Act. That's what we built PrivatAI for — now with the performance that production work demands.
Try it
Plans start at €30/month; the chat is the fastest way to a first impression. For IT service providers building AI into client projects: the API is OpenAI-compatible and a DPA is available — get in touch and we'll set you up personally.
PrivatAI (privatai.com) — AI chat & API from Germany. No content logs, no training on your data, GDPR-compliant.
