Artemis Security launches Orion-1 security model in private preview
Overview
Artemis Security has launched Orion-1, a model that investigates attacks on corporate networks and recommends or starts responses, in a private preview for select customers.
Customers set limits per action type on how much the model can decide on its own, and high-impact actions need human approval by default.
Artemis says Orion-1 scored 80.8 on attack reconstruction in its own Decision-Grade Readiness benchmark, against 58.3 for the next-best model, which the company reports as Claude Opus 5.5 or GPT-6 Sol. The benchmark is company-run and has not been independently verified. The Decoder also reports that frontier model guardrails have blocked defensive security analysis, and that Artemis has disclosed $70 million from seed and Series A rounds.
Written by AI from the articles below · updated Oct 14, 9:43 AM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- The DecoderNewsArtemis Security launches Orion-1, a security model for investigating attacks
AIArtemis Security has launched Orion-1, a model that investigates attacks on corporate networks and recommends or starts responses, and is in private preview. On the company's own Decision-Grade Readiness benchmark, Orion-1 outperformed Claude Opus 5.5 and GPT-6 Sol, scoring 80.8 on attack reconstruction versus 58.3 for the next-best model. The article also reports that frontier model guardrails have blocked defensive security analysis, and that Artemis has disclosed $70 million from seed and Series A rounds.
Heat trend
Not enough continuous observations to show a trend yet.