New proprietary AI models, WilBERT and Ventris, achieve 94% source code recovery accuracy, enabling organizations to automate binary analysis with the speed and reliability required for real-world security operations

LONDON, Sept. 02, 2026 (GLOBE NEWSWIRE) -- RevEng.AI, a cybersecurity company building the binary-native verification layer for the software supply chain, today announced the launch of Mega Bite, a suite of specialized AI models for binary analysis. This includes two new proprietary AI models, WilBERT and Ventris, which are purpose-built to analyze binary software without source code. Available within the RevEng.AI platform as an extension of the company’s signature BinNet technology, the models give security teams a novel way to detect malicious code, analyze malware, and verify software running in production, establishing a stronger foundation for AI-powered binary analysis.

Organizations today rely on more third-party, commercial, and AI-generated software than ever before. But by the time that software reaches production, it typically exists as a stripped binary, where much of the original source code context is no longer available. This makes it significantly difficult for security teams to understand what is inside the software, limiting their ability to identify vulnerabilities, investigate malicious behavior, and verify its overall integrity.

Traditional decompilers attempt to reconstruct that missing context through a complex set of rules and edge cases, but the process is inherently inexact and can produce pseudocode that is challenging for humans to interpret. General-purpose large language models also struggle to retrieve this because they were not designed for source code recovery and often hallucinate when analyzing stripped binaries. As a result, existing approaches require extensive manual analysis and do not provide the level of accuracy organizations need to confidently automate software security workflows.

WilBERT and Ventris were built specifically to solve this challenge by reliably recovering source code with the accuracy needed for real-world security workflows. Together, the models advance AI-powered supply chain security through distinct functionality and performance:

Specialized capabilities: WilBERT quickly identifies similar pieces of code across binary artifacts while Ventris explains and recovers semantically equivalent source code for each executable, enabling security teams to automate binary analysis with greater speed, assurance, and precision.
Extensive training: Trained on more than 50 trillion tokens of code-to-binary pairs collected from across the internet and built in a myriad of ways, both models learn to recognize the information critical to source code recovery and produce semantically correct versions of source code that avoid hallucinations.
High accuracy: The models achieved 94% accuracy on the HumanEval benchmark for source code recovery, compared to 48% for Anthropic's Fable model and 45% for OpenAI’s GPT-5.5.

"Software development is accelerating faster than security teams can manually inspect every application and dependency,” said James Patrick-Evans, PhD, Founder and CEO of RevEng.AI. “AI has the potential to fundamentally change how organizations can assess software risk, but only if it delivers results they can trust. Establishing that trust is foundational to RevEng.AI, with WilBERT and Ventris strengthening our platform’s ability to provide greater confidence in the software organizations build, buy, and deploy.”

With inference speeds of approximately 10,000 tokens per second, the models work together to automate binary analysis across existing security workflows at scale. By recovering the source code and context that is typically lost during compilation, WilBERT and Ventris enable security researchers and engineers, malware analysts, bug hunters, and incident response teams to analyze software before it reaches production and make more informed decisions about suspicious or unwanted code entering the software supply chain.

“As AI generates more of the software organizations rely on, knowing what is inside that software before trusting it becomes critical,” said Chris Steffen, Vice President of Research at Enterprise Management Associates. “That level of assurance becomes even harder to achieve with third-party code and complex dependency chains. By creating models designed specifically for binary analysis, RevEng.AI gives security teams greater visibility into what actually runs in production and a way to verify software without needing access to the original source code. This is something the industry has needed for a long time.”

To learn more about WilBERT and Ventris, visit RevEng.AI, read the technical blog, and watch this video for a deeper look at how the models provide reverse engineering quality analysis at machine speed.

About RevEng.AI
RevEng.AI is building the binary-native verification layer for software supply chain integrity. Founded in 2023, the company helps enterprises and defense organizations analyze compiled software—including closed-source and third-party artifacts—to understand what is actually inside the software they ship, buy, and run. By automating reverse engineering, binary analysis, and integrity verification, RevEng.AI helps customers detect hidden threats, reduce software supply chain risk, and make better trust decisions before deployment.

Media Contact
Sarah Ribacoff for RevEng.AI
sarah@aspironinfluence.com

This item of content is a press release from a third party provider who make no warranties or representations in connection with this content and therefore PA can make no warranties in respect of its use.

Tags

More Like This

RevEng.AI raises $15 million to verify the security and integrity of AI-generated software

Armis Launches First-of-Its-Kind Benchmark Report Warning of Critical Security Gaps in AI-Native Development

Binarly Launches Next-Generation Transparency Platform to Elevate Software Supply Chain Security

vLex Launches Major Upgrade to Vincent AI, Creating the World's Most Comprehensive AI Legal Research Assistant

LLMs Are Getting Smarter, But Not Safer: Veracode 2026 GenAI Code Security Report Finds AI-Generated Code Security Has Stalled at 56% Pass Rate

AI-Generated Code Poses Major Security Risks in Nearly Half of All Development Tasks, Veracode Research Reveals

PortSwigger Unveils Burp AI, Pioneering AI-Powered Web Application and API Security Testing

Announcing Veeva AI