Lyrie Agent reproduced 1,495 of 1,507 real-world vulnerabilities with working exploits in a single trial — the highest reported score on UC Berkeley's CyberGym Level 1 benchmark, above listed entries from Google, Anthropic, OpenAI and Huawei Cloud.
-- DUBAI, UAE, October 24, 2026 — Lyrie AI, the autonomous cybersecurity company behind Lyrie.ai, has achieved the highest reported score on CyberGym's Level 1 leaderboard: 99.2% success — 1,495 of 1,507 real-world vulnerabilities reproduced with working, executable proofs of concept in a single trial per instance. The result is dated September 15, 2026, and was checked against the public leaderboard on October 23, 2026.
CyberGym, developed by UC Berkeley researchers, tests whether AI agents can reproduce real software vulnerabilities with working exploits — not flag them, but actually trigger them. At Level 1, an agent receives a vulnerability description and an unpatched codebase spanning 188 real software projects, with one attempt per instance.
Lyrie Agent holds the world's No. 1 reported CyberGym Level 1 score: 99.2% exploit-reproduction success.
The official leaderboard data places Lyrie Agent above every listed entry from the world's major AI labs: Huawei Cloud (95.4%), xAI (91.3%), Anthropic (91.0%), Google (86.3%), OpenAI (85.6%), Tencent (84.8%) and Meta (59.0%). No listed entry on the public table scores above Lyrie's 99.2%.
"Reproducing a real vulnerability with a working proof of concept is the hardest evidence an autonomous security agent can produce," said Guy Sheetrit, founder and CEO of Lyrie AI. "It means reasoning and execution converging at machine speed — and that is exactly what our commercial platform brings to authorized testing and remediation."
The system combines AI-generated inputs, automated fuzzing and an iterative generator loop guided by execution feedback, then verifies that the resulting artifact actually triggers the fault. Lyrie's published technical report accounts for all 1,495 solved instances.
The commercial platform carries the same verify-first philosophy into production: OMEGA, an autonomous penetration-testing system that maps attack surfaces, discovers attack paths, tests exploits and chains vulnerabilities; and Lyrie Remedy, which analyzes findings, generates targeted patches, deploys fixes and verifies results while preserving original files. The wider platform includes web application defense, endpoint protection and security for AI-agent infrastructure.
Lyrie AI is now inviting venture capital firms, strategic investors, enterprise technology partners and potential acquirers to help scale product delivery, expand distribution and demonstrate customer outcomes across a wider market. The full analysis, verified scoreboard and benchmark disclosure are published at Lyrie Research.
About Lyrie AI
Lyrie AI (Lyrie.ai), built by OTT Cybersecurity LLC, develops cybersecurity technology spanning autonomous penetration testing, vulnerability scanning, remediation, web application defense and endpoint protection. Founder and inventor: Guy Sheetrit. HQ: Dubai, UAE.
Media Contact
Guy Sheetrit — Lyrie AI — [email protected] (investors) / [email protected] (partnerships) — lyrie.ai
Benchmark disclosure: Ranked No. 1 by reported score among CyberGym Level 1 single-trial, Agent-focused entries, checked October 23, 2026. The leaderboard labels the displayed metric "Reference success rate." Results are evaluated and submitted by individual teams; runs are stochastic, and modest score differences may not indicate meaningful capability gaps. Agent-focused and Model-focused entries are separate categories. CyberGym evaluates vulnerability reproduction; this result does not establish the performance of Lyrie's commercial protection or remediation products.
Contact Info:
Name: Guy Sheetrit
Email: Send Email
Organization: Lyrie AI
Phone: +1 (424) 666-9521
Website: https://lyrie.ai
Release ID: 89207246
In case of detection of errors, concerns, or irregularities in the content provided in this press release, or if there is a need for a press release takedown, we strongly encourage you to reach out promptly by contacting [email protected] (it is important to note that this email is the authorized channel for such matters, sending multiple emails to multiple addresses does not necessarily help expedite your request). Our efficient team will be at your disposal for immediate assistance within 8 hours – resolving identified issues diligently or guiding you through the removal process. We take great pride in delivering reliable and precise information to our valued readers.
