Kindo Launches WhiteRabbitNeo Version 3: The Next Chapter In Autonomous, Open Cybersecurity & Infrastructure AI

VENICE, Calif., May 15, 2025 /PRNewswire/ -- Kindo, the AI-native infrastructure and security automation company, today announced the launch of WhiteRabbitNeo V3, the latest version of its leading domain-specific cybersecurity and DevOps AI model. Purpose-built for offensive and defensive security, infrastructure automation, and agentic operations, WhiteRabbitNeo V3 brings unprecedented power and accessibility to security teams—from cloud-native enterprises to government environments.

WhiteRabbitNeo launched in December of 2023 is the world's first cybersecurity AI and already the most widely adopted open security AI model family, with over 140,000 downloads, 20,000 cybersecurity enthusiasts and engineers sign up at WhiteRabbitNeo.com every month and more than 1,400 enterprises use its models to tackle a wide variety of security challenges from enriching incident response work, checking code for security vulnerabilities and easily generating automation-ready security and infrastructure runbooks.

With V3, Kindo doubles down on its mission: to deliver autonomous infrastructure secured by AI-native agents that actually act—not just talk.

"In a world where security threats are increasing and enterprise systems are only getting more complex, static SOAR and chatbot wrappers just aren't enough," said Andy Manoske, VP of Product at Kindo. "WhiteRabbitNeo V3 is the first security AI designed from the ground up to take real action."

"We designed V3 to operate wherever enterprise workloads live—whether that's on a laptop, in the cloud, self-hosted, or fully air-gapped," said Bryan Vann, CTO of Kindo. "It's not just smarter, it's deployable, secure, and ready to operate autonomously at the edge of today's infrastructure."

Built for Action, Trained for Security

WhiteRabbitNeo V3 has been upgraded with new domain training in web security, infrastructure-as-code, and vulnerability research. It now features improved tool calling, long-context stability, and better performance across multi-step tasks. The result is a more autonomous model that can handle tasks like exploit simulation, CVE validation, and live system remediation faster and with greater reliability.

Unlike censored commercial LLMs, WhiteRabbitNeo is open-weight, uncensored, and deployable in fully private, air-gapped environments—a critical requirement for sensitive enterprise and government workloads.

Small Model, Big Results

Despite its compact size at 7B parameters, The open version of WhiteRabbitNeo V3 has demonstrated benchmark performance on par with models 10X its size. In Cybench, a real-world hard for AI to solve benchmark designed to test AI models on Capture-The-Flag cybersecurity challenges, WhiteRabbitNeo V3 solved 5% of tasks autonomously—matching much larger systems while running on a single GPU. This level of efficiency is currently unmatched. The AI industry's best performing frontier foundation model achieves just 22.5% on CyBench and is 60-100X more resource intensive than WhiteRabbitNeo V3 while still not being usable reliably in autonomous security agents due to censorship issues breaking workflows nor can they be ran privately which is an essential need for most sensitive security and infrastructure work.

This makes WhiteRabbitNeo V3 the most efficient, deployable, and cost-effective security and DevOps model available today. Kindo engineers have even run the model performantly on better equipped Apple MacBooks, showcasing its portability and affordable accessibility for developers, system engineers, and defenders everywhere.

Built into Kindo's AI-Native Platform

At the core of Kindo's autonomous AI platform, WhiteRabbitNeo V3 powers real-time automation across SecOps, DevOps and ITOps workflows. Users can describe their intent in natural language, and the AI agent executes on it—whether that means identifying a misconfigured firewall, understanding the impact of a possible compromise, simulating a zero-day exploit, troubleshooting complex authorization issues across a stack of cloud services or generating infrastructure code with embedded security and compliance policies.

Security and infrastructure platform teams using Kindo with WhiteRabbitNeo V3 gain:

• Faster incident response and triage
• Context-rich remediation across complex stacks
• End-to-end infrastructure instrumentation with minimal manual toil
• Deeper understanding of security and infrastructure reliability risks
• A model that will work with AI agents on complex security and infrastructure risks without having to worry about censorship breaking autonomous workflows

Open and Enterprise-Ready

WhiteRabbitNeo is available in both open and commercial form. The open-weight version provides the security community with an auditable, powerful tool for hands-on security research and autonomous defense. Meanwhile, enterprise customers gain access to a fully integrated, production-ready, and supported inside Kindo's secure AI orchestration environment.

Kindo's long-term vision: a future where enterprise infrastructure is greatly simplified, deployed, managed, and secured by autonomous AI-native agents that understand your intent or the intent of a threat to your business—and when needed take action.

References

Try Whiterabbitneo V3 for free: https://app.whiterabbitneo.com/
Learn more about Whiterabbitneo: https://www.whiterabbitneo.com/
Learn more about Kindo: https://www.kindo.ai/

About Kindo
Kindo is the platform for AI-native security and autonomous infrastructure. Built from the ground up to realize intent through intelligent agents, Kindo powers mission-critical automation across SecOps, DevOps and ITOps—so enterprise teams can do more with less, faster, and more securely.

SOURCE Kindo