Frequently asked questions

Chancellor models are not large language models. A Large Action Model (LAM) — also called an Action Learning Model — is trained end-to-end to plan, decide, and execute multi-step actions: calling tools, operating software, driving robots. Text is one of its instruments, not its whole world. That's why our playground is called the Actionfield.

The Actionfield is our interactive playground for talking to Chancellor models — picking a model, tuning parameters, and inspecting ChancellorTrace matches against the open training data. In the current preview, responses are simulated so you can explore the interface before the hosted models go live.

Not yet. This preview simulates responses in your browser — no data leaves your device. When hosted inference launches, the interface will say so explicitly and our Privacy Policy will be updated first.

Not yet — the Chancellor family launches one model a month this autumn: Chancellor-1 70B on September 9, Chancellor-1 8B on October 10, and Chancellor-VLM on November 11, 2026. On each launch day everything ships at once under Apache-2.0: weights, training data audits, training code, and evaluation suites. Write to launch@ichancellors.com to get notified.

Weights, data recipe, training code, intermediate checkpoints, evaluation suite, and known failure modes — all public. If we learned it, you can download it. Openness is our method, not our marketing.

Because our training corpus is public, we can show you documents from the training data that have exact text matches with a model's response. Click “Show ChancellorTrace” under any completed response in the Actionfield.

Membership is by request: you send us a short email, a human reads it, and we approve joins personally. Head to the Community page — the request form there opens a pre-filled email to us.

No. In this preview, conversations and thread history live in your browser's local storage only. Clearing your browser storage for this site erases them completely.

Safety findings go to safety@ichancellors.com — we respond within two business days and credit reporters in our published audits if they wish. Everything else: press@ichancellors.com for media, talent@ichancellors.com for careers.

Yes — across Research AI, Physical AI, Quantum, Social Science, and Operations. See the Careers page: 18 roles are open, and exceptional researchers can pitch their own.