LaunchPolicyNovember 15, 2026

Axiomic Labs releases X3Guard-150M prompt guard model

Read original source →huggingface.co

The 150M-parameter guard, trained on Axiomic's GPT-X3 base model, flags a prompt as Harmful and Adversarial in a single pass. The lab claims it outperforms models many times its size and can run locally.

1 source

More stories today

Open the live feed