Reduced compute costs by 90% without disrupting development timelines
Achieved a 72% success rate in verifying and exploiting vulnerabilities
Scaling vulnerability research to test against up to 5x+ more potential exploits
Quantro Security moves its AI to Google Cloud and saves 90% on compute costs to scale vulnerability and exploit research.
Quantro Security moves its AI to Google Cloud and saves 90% on compute costs to scale vulnerability and exploit research.
Keeping pace with AI threats
using a diverse set of models
Artificial intelligence has rapidly and dramatically altered cyber offense and defense. In the middle of this revolution, Quantro Security develops novel, agentic AI-powered approaches to vulnerability detection and exposure management. Customers trust Quantro Security to help them find security risks and protect against exploitation. To power this cutting-edge cybersecurity research, Quantro’s Exploit Harness reasons, evaluates, and identifies vulnerabilities 24 hours a day, seven days a week.
The team relied on a wide range of models from different providers to power these demanding workloads, but that resulted in several operational and financial bottlenecks. Because the company was tied to standard API pricing with limited options for cost control, compute expenses ballooned.
We felt so limited in our ability to control costs. We started to explore our options and realized we didn’t have to move our entire workload to one platform. We could just shift some of it to test, and that opened up so many new possibilities.
Mehul Revankar
Co-founder and CPO, Quantro Security
Some of these providers were also inconsistent with unexplained timeouts and restrictive usage limits that disrupted their development pipelines. When an API limit was unexpectedly reached, processes would halt and waste tokens. This meant the team had to restart some complex experiments from scratch. “We felt so limited in our ability to control costs,” says Mehul Revankar, Co-founder and CPO at Quantro Security. “We started to explore our options and realized we didn’t have to move our entire workload to one platform. We could just shift some of it to test, and that opened up so many new possibilities.”
After being limited in fine-tuning models and managing a closed environment that felt particularly vulnerable to data leakage risks, the company decided to work with Google Cloud to scale their cutting-edge research.
Scaling secure, multi-model research models
on Google Cloud
Through Model Garden on Gemini Enterprise, the team gained the flexibility to experiment with a wide variety of open-source models outside of their previous closed ecosystem. After extensive testing to find the highest value output for their specific use cases, Quantro selected a combination of models, including Kimi, GLM 5.2, Gemma, and DeepSeek Pro to power their continuous evaluation.
Instead of relying on uncontrollable API pricing, Quantro deployed these models on reserved compute instances. This dedicated, high-performance infrastructure ensured it had the necessary compute capacity to run their complex experiments 24 hours a day.
Google Cloud was already the cornerstone of our tech stack. So consolidating into even fewer platforms not only improves our model work but also improves our ability to monitor, observe, and respond to any problems.
Sasan Padidar
Co-founder and CEO, Quantro Security
This approach gave Quantro the flexibility it needed to deliver its newest, large-scale research project: The Economics of Vulnerability Exploitation with AI. The diverse, open-source model options now manage 90% of the workloads in Quantro’s pipeline. For the remaining 10% of highly specialized edge cases, the team could easily pivot back to other frontier models, so they can shape their AI infrastructure to the task at hand. “Google Cloud was already the cornerstone of our tech stack,” shares Sasan Padidar, Co-founder and CEO. “So consolidating into even fewer platforms not only improves our model work but also improves our ability to monitor, observe, and respond to any problems.”
Additionally, moving to Google Cloud provided a more secure environment where Quantro retained 100% ownership of its proprietary data. With that ownership, it can more freely fine-tune models tailored specifically to the team’s vulnerability research.
Improving global cybersecurity best practices
with AI-tested security data
With its dynamic, multimodal solutions running on Google Cloud, Quantro has accelerated its groundbreaking vulnerability detection and exploit research. The company has successfully reduced its compute costs by 90%, while continuing to improve its Exploit Harness’s capabilities. The team achieved this massive cost reduction while maintaining an impressive 72% success rate in verifying and autonomously exploiting vulnerabilities.
Quantro’s new AI infrastructure also allows them to analyze thousands of common vulnerabilities and exposures (CVEs) in record time. With a more stable infrastructure and more predictable costs, Quantro is no longer held back by arbitrary API limits. The company now plans to scale its research from testing against 3k CVEs to up to 20k by the end of 2026.
We need to fight AI with AI, and we’re excited to be a part of a generation of companies that are building AI-native technologies for defenders. From our point of view, we’re well positioned to do that through our relationship with the Google for Startups team.
Mehul Revankar
Co-founder and CPO, Quantro Security
While the company continues to scale, it plans to experiment with changing trends in AI model adoption. “Right now, companies defer to frontier models, but over time, I think we’ll see a push toward owning models and owning hardware to gain as much control as possible,” Revankar adds. He also trusts that Quantro Security’s approach to building on the open-source solutions available through Google will give the team the options they need to combat AI threats.
“We need to fight AI with AI, and we’re excited to be a part of a generation of companies that are building AI-native technologies for defenders,” Revankar adds. “From our point of view, we’re well positioned to do that through our relationship with the Google for Startups team, and the larger Google Cloud suite.”
Quantro Security is an AI-native cybersecurity company whose agentic workforce safely exploits new vulnerabilities within minutes, so customers can patch them before attackers do.
Industry: Technology
Location: United States
Products: Google Cloud, Gemini, Gemini Enterprise
