
OBLITERATUS
Study and modify model refusal behavior, then measure what changes.
Open tools, models, and infrastructure for exploring what intelligence can do.

Study and modify model refusal behavior, then measure what changes.

We build open-weight models to expand what red teamers and white hats can test, discover and experiment with—helping them harden systems and uncover blind spots in our collective understanding.

Open chat for comparing models and exploring their responses.

A multi-agent framework for authorized security testing.

An AI interface for Flipper Zero, with voice, text and vision.

A workbench for codes, ciphers, alphabets and promptcraft.

Join the mission
Tools, model experiments, robotics, repo maintenance, and research infrastructure.
Join Engineering