Continuum AI Introduces OrcaCyber Zero 1.5, a Frontier Cybersecurity Reasoning Model

Updated model reports 95.8% on CVE-Bench and 100% on Cybench, advancing autonomous vulnerability discovery, exploit validation and remediation.

SAN FRANCISCO, Oct. 12, 2026 /PRNewswire/ -- Continuum AI announced OrcaCyber Zero 1.5, its latest frontier cybersecurity reasoning model built to discover critical vulnerabilities, investigate multi-stage attack paths and validate real-world exploitability. Featuring a 1 million-token context window, native tool calling and extended reasoning, the updated model is designed for complex, autonomous security investigations across large codebases and interconnected systems. Security teams can apply its reasoning to complex repository structures, extensive security logs and lengthy investigation histories without separating evidence into isolated sessions.

BENCHMARK PERFORMANCE
In company-reported evaluations, OrcaCyber Zero 1.5 achieved 95.8% on CVE-Bench (23 of 24 evaluable tasks), 100% on Cybench (39 of 39 challenges, unrestricted) and 93.9% on HumanEval+. Its predecessor, OrcaCyber Zero 1.0, powered an autonomous harness achieving over 98% pass@1 on CyberGym, ranking among the top three globally according to Continuum AI. Zero 1.0 has also supported security work for major software companies, government organizations, blockchain networks and cybersecurity firms.

SIX CONTROLLED SECURITY DEMONSTRATIONS

• Administrator takeover: Exploited SQL injection to bypass authentication, obtain administrator access and expose user records.
• Sensitive file access: Located restricted files and demonstrated an access-control bypass.
• Cross-site scripting: Validated JavaScript execution in a real browser.
• Business-logic exploitation: Manipulated checkout quantities to create a negative-value order.
• AI-versus-AI prompt injection: Extracted hidden instructions from another AI assistant in a controlled test.
• Autonomous remediation: Discovered a vulnerability, reproduced it, generated a patch and verified the fix through tests.

FROM ALERTS TO ACTIONABLE CYBER DEFENSE
OrcaCyber is designed to prioritize reproducible evidence over alert volume. By combining vulnerability discovery, exploit validation, root-cause analysis and tested remediation, it aims to help defenders focus on genuine, high-impact weaknesses while reducing false positives. The broader objective is to move security operations beyond passive scanning toward autonomous investigation and defense, with human oversight for consequential actions. This approach connects individual findings into coherent attack narratives, helping analysts distinguish theoretical weaknesses from validated risk and prioritize fixes backed by reproducible tests.

Continuum AI also plans to release an even more capable open-weight cybersecurity model next week, designed for local deployment. The release will aim to make frontier cybersecurity intelligence more accessible while giving organizations direct control over sensitive security workloads and infrastructure.

AVAILABILITY AND COMPANY INFORMATION
Security researchers and enterprise teams can explore OrcaCyber Zero 1.5 on OrcaRouter and request model access. Continuum AI develops frontier cybersecurity models and secure AI infrastructure, including the privacy-first OrcaRouter AI gateway. Learn more about the company at continuum01.ai.

Media Contact:
Chris Zhang
650.609.7501
[email protected]

SOURCE Continuum AI