You can track all your projects, tasks, usage, and model calls in one place on ORGN Studio
Full visibility into your confidential AI usage. No black boxes.
Stop guessing. Start tracking → https://t.co/1XM7vhLfhQ
GLM 5.2 is LIVE in ORGN
Now with hardware-level privacy:
- zero data retention by default
- deployed on confidential computing chips
- verifiable hardware attestations for every request
Use the world's best models without handing over your data or IP.
AI chats don't stay private
Recent events:
• Deepseek: +1M chat logs exposed
• Grok: chats on Google
• AI companion app: 43M messages leaked
ORGN Gateway makes your AI traffic hardware-private.
200+ Vetted, zero-retention + TEE models. 1 API
Get your API Key now → https://t.co/4v8ZahitOv
@orgn_official Oh wow, I completely missed this. I do understand the willingness to prevent Fable for being used for evil, but communicating that your model is so capable that it is dangerous for cybersec is bound to tip some scales for sure 😅
We just watched export controls move from GPUs to HTTPS
Immediately implications:
• You cannot enforce this kind of control long‑term without strong ID verification on model usage.
• Anonymous frontier access is now politically radioactive
This, of course, raises some important concerns.
If frontier model research and development is centralised in the hands of a few, US-based companies then access to those models becomes a matter of geopolitics, and a single directive can decide which passports get to use them and which ones don’t
That's why open source is so immensely important, and why we wholeheartedly support it.
You can run it on your own hardware, or like we do at ORGN, wrap it in TEEs for verifiable confidentiality. No government can suddenly revoke your API key. Access isn’t gated by nationality, and the whole field can keep iterating, for everyone, instead of waiting to see who Washington decides gets to use the latest model this week.
Quick framework for choosing your AI model in ORGN:
Routine work? → ZDR model. No data retained, fast, hundreds of options.
Sensitive work? →TEE model. Hardware isolation, attestation, nothing leaves the trust domain.
You have a spare kidney? → Fable 5
Claude Fable 5 is here!
Available now in ORGN Gateway and natively in ORGN CDE.
Please note: this is the only model we offer that isn't ZDR. Every other model is. The choice is yours.
Happy coding!
How to code with AI on sensitive tasks with full confidentiality
This thread will run you through the basics of confidential worktrees
If you work on sensitive code, it might be the most important concept in your workflow.
Let's go 👇🧵
Our docs just got a full redesign!
One unified space to learn about every part of the ORGN stack: CDE, Studio, Gateway, and Scanner.
If you've been confused about ho.w the pieces fit together, start here 👇
Local or fully confidential. Your call
Work with your local files to stay on your machine, or hop into confidential cloud projects to work across tasks and launch Intel TDX worktrees
We built Origin so a compromised sandbox can't reach your local machine.
A targetted attack has no way of accessing your filesystem, your credentials, or secrets, unless you explicitly inject them as context into ORGN
This puts the control in YOUR hands. You decide what is seen. You decide what to expose and where.
Ok but can Claude Code give you cryptographic proof that no one saw your code, trained AI on it, or sold it to third parties during your coding session?
Yeah, didn't think so
The ORGN CDE Desktop app is now also available for Windows users
Same coding agent performance as Cursor, Claude code or codex at a fraction of the price.
Same experience. No friction.
With confidentiality built into the architecture down to the silicon.
ORGN is evolving.
We're going from a confidential AI IDE to a full confidential agentic stack: infra, agents, apps
Confidentiality is by design, down to the silicon.
ORGN CDE is now just one piece of the system.
OLLM is becoming ORGN Gateway.
OLLM was the confidential AI gateway. Now it's part of ORGN: the full confidential agentic stack.
Studio. Gateway. CDE. Agents. Sandboxes.
Same power. Bigger vision.
GitHub was hacked through a malicious VS Code extension, and it wasn’t a sophisticated attack...
It came down to trust in the IDE, and that detail matters more than you think if you are a developer.
Here’s what you can learn from it:
Tomorrow we ship something everyone has been asking for.
No new flow, no learning curve. Just the dev experience you already know and love
with something you can't get anywhere else
See you tomorrow.