Cisco's Small Open-Source Models vs. GPT-5.5: Finding Vulnerabilities at a Fraction of the Cost

In July 2026, Cisco released two compact open-source AI models for cybersecurity. By the company's own tests, they find roughly 150 times more vulnerabilities per dollar spent than large general-purpose AI agents.

What exactly did Cisco release?

Cisco has open-sourced two small models built for a single task: finding vulnerabilities in code and systems. Unlike general-purpose giants such as OpenAI's GPT-5.5, these are narrowly specialized models. They have fewer parameters, cost less to run, and are openly available, so they can be deployed on your own infrastructure.

The move fits a broader trend: instead of one giant model for everything, companies increasingly pick a small model for a specific job. Cisco applies that logic to security, a field where both the cost of scanning at scale and control over data matter.

• Two separate open-source models, both for cybersecurity
• Specialization: vulnerability detection
• Claimed advantage: about 150x more vulnerabilities found per dollar compared with large AI agents
• Comparison benchmark in the announcement: OpenAI's GPT-5.5
• Open license: the models can be run locally, without sending code to external services

Why is it 150 times cheaper?

Model size drives the economics. A large agent built on GPT-5.5 spends money on expensive inference with every run, and scanning a large codebase takes thousands of runs. A small model costs orders of magnitude less per call for the same task. That per-call price gap, multiplied by the volume of checks, produces the claimed difference of roughly 150x in vulnerabilities found per dollar.

An important caveat: the 150x figure comes from Cisco's own internal tests, not from an independent benchmark. In its brief announcement, the company did not disclose the exact measurement methodology or the datasets used to compare the models. Independent testing on public codebases will show what the real gap is.

By

Cisco's own tests, the compact models find roughly 150 times more vulnerabilities per dollar than large AI agents, The Decoder reports, citing the company's announcement.

What changes for security teams?

For cybersecurity teams, the key point is not just price but openness. A proprietary agent requires sending code fragments to someone else's cloud, which is often unacceptable for security work: source code and discovered vulnerabilities are sensitive data. Cisco's open-source models can be deployed inside the perimeter, so code can be scanned without ever leaving it.

The second shift is how often checks can run. A small model can realistically run continuously in a CI/CD pipeline on every commit, while an expensive GPT-5.5-based agent is more often run selectively because of its cost. With a price gap of roughly 150x, constantly scanning every change becomes economically justified rather than an occasional luxury.

Finally, a major industry player stands behind the release. Cisco is one of the leading suppliers of networking and security equipment, and the fact that it is open-sourcing specialized models raises the odds that real SOC and DevSecOps workflows will adopt them rather than leave them as an experiment.

What this means

Cisco is betting that in cybersecurity, narrow specialization and open access matter more than raw general-purpose power: by its tests, two small models are cheaper and more private than a large agent, and the price gap is measured not in percent but in hundreds of times. There is no independent verification of these numbers yet, and that is what will show whether the 150x claim holds up against real codebases.