Claude Mythos, run by LESIS
Your path to Mythos.
Anthropic built a model that finds and exploits vulnerabilities on its own, then decided the public should not have it. Claude Mythos is reserved for the companies behind the world's critical software and a small group of verified security organizations.
LESIS is one of those organizations, verified by Anthropic through its Cyber Verification Program. We run Mythos on your code and systems, prove what it finds and hand your team the fixes.
On the record
- Thousandsof high-severity zero-days found in a few weeks, across every major operating system and web browser.
- 27 yearsthat a flaw letting anyone crash an OpenBSD machine went unnoticed, until Mythos read the code.
- 5 milliontimes automated testing ran the FFmpeg line that hid a 16-year-old bug Mythos found.
- 83.1%on CyberGym vulnerability reproduction, against 66.6% for Claude Opus 4.6.
Why it matters
Every line of code you have ever shipped is about to be read by something that never gets tired.
In a few weeks, Mythos found thousands of high-severity vulnerabilities in every major operating system and web browser. A flaw in OpenBSD, 27 years old, that let anyone crash a machine just by connecting to it. A bug in FFmpeg, 16 years old, on a line automated testing had run five million times. Linux kernel bugs chained from an ordinary user to full control of the machine. Nearly all of it with no human steering.
None of those bugs were new. What changed is the cost of finding them. Capability like this does not stay in one lab for long, and once attackers have it, the time between a flaw and its exploit shrinks to hours. The organizations that come through are the ones whose code a model like this read first, on their side.
How LESIS runs it
You get the findings.
We carry the model.
Access, scope, compute and the engineers who check every result come from LESIS. What reaches your team is a short list of proven vulnerabilities, each with a way to fix it.
Scope and authorization
We agree on what is in scope, from repositories to applications and infrastructure, with the written authorization these programs require. Nothing runs outside it.
Found by AI, proven by LESIS
The model reads your code and systems the way an attacker would. LESIS engineers reproduce every result before it counts, so you get vulnerabilities, not suspicions.
Into your record
Every proven finding lands in the LESIS platform with its evidence, severity, an owner and a deadline, next to the rest of your program.
Fixed, then retested
When your team ships the fix, we run the check again and close the finding only when the hole is gone.
Access
Why you can't just sign up.
Anthropic does not sell Mythos to the public. It gives access to the partners of Project Glasswing, the companies that build and maintain critical software, and to vetted security organizations through its Cyber Verification Program, under rules on how the model is used. LESIS has been through that verification, so you get what Mythos finds without applying, waiting for approval or building a team to run it.
Sources
Claude Mythos is developed by Anthropic and Daybreak is a program of OpenAI. Their names are used here only to describe the access LESIS provides. Figures are as published by Anthropic and OpenAI or reported by the press.
Let Mythos read your code before someone else's model does.
Tell us what you run. We will tell you where Mythos should look first and how soon it can start.
