Artemis Security released Orion-1 on October 14, 2026, saying it investigates and responds to cyberattacks on corporate networks autonomously. It is the company's first cybersecurity-focused model since its stealth launch in April.
Artemis reported an 80.8 score in attack reconstruction, measured on its Decision-Grade Readiness benchmark. That compares with 58.3 for the next-best model, GPT-6 Sol.
Orion-1 is built on a custom architecture and targets corporate network security. Availability begins with a private preview for select customers on the Artemis platform, initially for enterprise clients.
"Orion-1 processes log data from auth services, cloud setups, corporate endpoints, SaaS apps, and network gear, then connects individual signals to tie them back to a single attacker," said Jonathan Kemper, Artemis Security's founder. The model assigns a confidence score, locks in a verdict, and recommends or kicks off a response.
The announcement follows Artemis' exit from stealth in April, disclosing $70 million from seed and Series A rounds. Artemis said the model is designed to close a gap created by general-purpose AI models that are becoming a growing problem for security teams.
Artemis did not say whether Orion-1 is built on an existing model from another provider, and it remains unclear how the company's fine-tuning process works. The company said it fine-tuned Orion-1 on tasks drawn from its own security operations, including spotting suspicious patterns in log data and containing live attacks.
Source: thedecoder
