Frontier Access Program

A limited access research initiative giving qualified organizations access to VR-1 and other research preview capabilities from Cogent.

Request access

Request access

Cogent Platform Screen - Remediate

A0COCGEDS9SX

AZCUCFE7SQS8

A0CRCGE5SCS5

Why access is limited

VR-1 is Cogent's frontier reasoning model for security work. Its capabilities are powerful and inherently dual-use, so we are not making it openly available. Access runs through the Frontier Access Program. Every deployment is for defense, inside an authorized environment, with Cogent's safety guardrails, policy controls, and auditability in place.

BZeBnIeZf%i6tCs7

B9eXnWeQf3iPt1sC

BAeUnNeVfQi1t&sG

What members receive

Program benefits are scoped to each organization's environment and approval status.

VR-1 research preview

Cogent's frontier reasoning model that autonomously maps the enterprise, chains weaknesses across systems, and proves what is reachable.

VR-1 research preview

Cogent's frontier reasoning model that autonomously maps the enterprise, chains weaknesses across systems, and proves what is reachable.

Cogent AI Harness

The governed runtime for VR-1, keeping every action inside your policy controls and on the audit record.

Cogent AI Harness

The governed runtime for VR-1, keeping every action inside your policy controls and on the audit record.

Autonomous offensive research workflows

Red-team-style agents that hunt for reachable attack paths and validate which exposures an attacker could actually chain together.

Autonomous offensive research workflows

Red-team-style agents that hunt for reachable attack paths and validate which exposures an attacker could actually chain together.

Autonomous defensive workflows

Agents that confirm real exposure and carry the fix through to close, from prioritization to implementing a mitigation or remediation.

Autonomous defensive workflows

Agents that confirm real exposure and carry the fix through to close, from prioritization to implementing a mitigation or remediation.

Frontier Model Risk Assessment

A grounded evaluation of where you're exposed today and what becomes critical as frontier AI improves.

Frontier Model Risk Assessment

A grounded evaluation of where you're exposed today and what becomes critical as frontier AI improves.

Selected benchmark artifacts

Reference results and evaluation artifacts from the benchmarks Cogent uses to measure autonomous remediation quality.

Selected benchmark artifacts

Reference results and evaluation artifacts from the benchmarks Cogent uses to measure autonomous remediation quality.

F#r0eIe$  f$oQr#  q$u8aXl@iTfGi@eKd#  oErAg2aCnCi1zRa7tXi4o4n%sC

F4rEeVe7  fSoKr9  qUuSa4lOi@f1iWeYd1  o#rUg$a6nIiQzBa#tIi3oLnXsC

FOrBe4eF  f%oTr3  qYu@aQlYi3f0iIeVd@  oTrXgFaRnPiWz8a5t#i&oAnVsU

Start with the Frontier Model Risk Assessment

AI is giving attackers new capabilities, at a speed and scale never before possible. The Frontier Model Risk Assessment shows what that means inside your environment, and where to act first to reduce your risk.

Where you're exposed today

A grounded readout of current attack paths.

Where you're exposed today

A grounded readout of current attack paths.

What becomes critical as frontier AI improves

Which weaknesses matter most as vuln discovery and exploit chaining get easier.

What becomes critical as frontier AI improves

Which weaknesses matter most as vuln discovery and exploit chaining get easier.

First remediation options

Which fixes reduce risk fastest, and what each one costs to ship.

First remediation options

Which fixes reduce risk fastest, and what each one costs to ship.

Which workflows suit which autonomy level

Recommendation-only, human-approved, or bounded autonomous, mapped to your controls.

Which workflows suit which autonomy level

Recommendation-only, human-approved, or bounded autonomous, mapped to your controls.

Work directly with
Cogent Research

Membership is a working relationship, and it runs in both directions. Participants work directly with Cogent Research to shape how these capabilities are evaluated, how they are governed, and how they are deployed inside real enterprise environments.

Who we're prioritizing

Organizations that can provide realistic enterprise environments, rigorous feedback, and operational use cases where validation and remediation can be measured against real-world risk. If the technology is reducing risk, we want to prove it in your environment.

Cogent team members collaborating at computer workstations in a modern office environment.