Skip to content

An editorial team and a community building with AI

← All articles

News · OpenAI · 2 min read

Event date: · Published:

OpenAI launched GPT-6 Cyber for authorized vulnerability research through Daybreak

On November 5, 2026, OpenAI released GPT-6 Cyber, a cybersecurity agent model trained to stay within authorized scope, with access set up through Daybreak. The preview model it replaced, gpt-5.5-cyber-preview, was set for removal on February 1, 2027.

On November 5, 2026, OpenAI released GPT-6 Cyber, a cybersecurity agent model that the company said was trained to stay within authorized scope. OpenAI described the release in its GPT-6 Cyber announcement, and access was set up through Daybreak.

What the release covered

  • Purpose: investigating vulnerabilities, reproducing them in an authorized test environment and verifying fixes.
  • Access: set up through Daybreak.
  • Scope: OpenAI said the model was trained to stay within authorized scope.
  • API pricing: $20 per 1M input tokens, $2 for cached input and $100 for output.
  • Cache writes: $25 per 1M tokens for 30 minutes and $50 per 1M tokens for 12 hours.
  • Deprecation: gpt-5.5-cyber-preview was set to be removed on February 1, 2027, according to a deprecation notice dated November 5, 2026.

What it means for people building with AI

The release made offensive security capability of the GPT-6 generation available to vetted defenders, with scope adherence as its main safety claim. The workflow OpenAI described was to investigate a vulnerability, reproduce it in an authorized test environment and verify the fix. The preview model gpt-5.5-cyber-preview, which GPT-6 Cyber replaced, was set for removal on February 1, 2027, giving teams that used it a fixed date to plan around.

Other OpenAI sources

OpenAI also listed GPT-6 Cyber in its API changelog on November 5, 2026, and published the removal notice for the preview model on its API deprecations page.


Sources

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.

OpenAI launched GPT-6 Cyber for authorized vulnerability research through Daybreak | vibecoding.inc