Antares scans five hundred code repositories for under a dollar. That speed is precisely why Cisco is screening who gets to run it before anyone downloads a copy.

Cisco released Antares-350M and Antares-1B on Hugging Face this week, two small language models Cisco Foundation AI trained for one job: finding which files in a codebase contain a known vulnerability. Both carry open weights. Both also require a defender-verification check before the download completes, and Cisco says it worked with United States government agencies on how the models should ship.

The Cost Case Holds Up

Software teams already know the expensive part of vulnerability response. A public advisory names a weakness class. Finding which files in a million-line repository contain it requires an analyst to trace call paths, compare naming conventions, and rule out false leads by hand. Antares automates that search. Given a vulnerability category and read-only terminal access to a repository, it searches, reads candidate files, and narrows toward a ranked list of likely culprits, the way an investigator works a lead rather than the way a chatbot answers a question.

On Cisco's new Vulnerability Localization Benchmark, Antares-1B reaches a File F1 score of 0.209, ahead of Z.ai's 753 billion parameter GLM-5.2 at 0.186 (MarkTechPost, 2026). OpenAI's GPT-5.5 still leads at 0.229, and costs accordingly: roughly $141 to sweep the same 500 repositories that Antares clears for under a dollar on a single H100 (MarkTechPost, 2026; Axios, 2026). Traditional static analysis tools trail all of them. Semgrep scores 0.086, CodeQL 0.023 (MarkTechPost, 2026).

That gap is the real story, and it holds regardless of what happens next in this piece. A model one two-thousandth the size of GLM-5.2 is finding more of the right files, faster, for less money, and it can run inside a customer's own environment instead of sending proprietary source code to a third party. For a university security team or a public-sector agency priced out of frontier-model token budgets, that difference decides whether continuous scanning happens at all.

Efficiency Is a Dual-Use Problem

A model that scans a repository for a known weakness in fifteen minutes for defenders can do the same search for attackers. The gap between disclosure and exploitation already runs in minutes for some vulnerability classes. A tool that speeds up the defender's search speeds up the attacker's search by the same margin, unless someone controls who gets to run it.

Cisco vetted the release with United States government agencies before shipping it and built a verification process to confirm downloaders as cyber defenders, according to Cisco senior vice president DJ Sampath (Axios, 2026). That is a considered response to a real risk, not an afterthought. It also means the university lab or public-sector team Cisco names as a target beneficiary has to clear that check before running the model at all, and clearing it takes time the smaller, better-resourced buyer may not need to spend.

Cisco has also not released Antares-3B, the version that scores closest to GPT-5.5 at 0.223 File F1. Cisco says that model will go into its own security products instead. Holding the strongest variant back while the smaller ones prove themselves in the field is consistent with the same caution behind the screening. It also means the model Cisco benchmarks against GPT-5.5 in its own materials is not the one available to download today.

"You really don't need a private jet to go to your corner store." — DJ Sampath, Cisco senior vice president and general manager of AI software and platform (Axios, 2026)

A Different Answer to the Same Problem

Capital One released its own agentic security tool five days earlier, on July 16. VulnHunter reasons from an attacker's perspective, maps out exploitable paths in source code, and proposes fixes, rather than only localizing known weaknesses. Capital One put it on GitHub under an Apache 2.0 license with no download screening at all.

VulnHunter runs on Anthropic's Claude Opus 4.8 through Claude Code, and Capital One manages the same dual-use risk Cisco is managing differently. Instead of gating who can download the tool, misuse gets caught at the model layer, through Anthropic's Cyber Verification Program screening requests in real time. Two vendors, five days apart, the same underlying tension, and two different places to put the control: Cisco at the download, Capital One at the point of use.

What the Benchmark Doesn't Show Yet

Cisco's benchmark runs two phases. Phase A hands the model a vulnerable snapshot and scores how many of the right files it finds. Phase B hands the model a patched snapshot, the fix already applied, and checks whether it still raises an alarm on clean code.

Cisco published Phase A results for every model it tested. Phase B results are not out yet for any of them.

That gap matters more than the headline F1 score for anyone planning to gate a continuous integration pipeline on this model's output. A vulnerability scanner that flags clean code at scale trains developers to ignore it within a month. Without a published true negative rate, a security team adopting Antares today is running its own experiment on false-alarm tolerance, with no benchmark yet to check it against.

Where Antares Sits in the Stack

Antares sits in a different layer of Cisco's security buildout than the identity work covered here this spring. Astrix, Galileo, and WideField Security all fed into Splunk's data plane, assembling a discovery, evaluation, and session-evidence stack for non-human identities in eight weeks. Antares operates upstream of that, at the point where a vulnerability advisory meets a codebase, before an agent or a credential enters the picture at all.

The staged rollout fits a pattern from those acquisitions too. Cisco has generally kept its most capable engines inside its own commercial products while open-sourcing the components built to earn trust and adoption around them. Antares-3B moving into Cisco's own security line while the smaller models validate in the open follows that same logic, and it is a defensible one for a dual-use capability still building a track record.

The practical question for a buyer is not whether Cisco's caution is justified. It is whether the smaller models perform well enough on their own, without Antares-3B and without published false-alarm data, to be worth building a workflow around this year.