Frontier-Level Red-Teaming From a Self-Hostable Open-Weight Model: Testing NVIDIA Nemotron 3.5 Super

Orca Security · Project update · October 15, 2026

Orca Security published research on NVIDIA Nemotron 3.5 Super, exploring how frontier-level red-teaming can be performed using self-hostable open-weight models to test the security of AI agents.

Read the research findings

AntSpace’s summary of Orca Security’s own announcement: read the original.

More updates from Orca Security