OpenAI puts GPT-5.6-Cyber behind a new Daybreak Red tier
GPT-5.6-Cyber is built for authorized vulnerability research, exploit validation, and defensive testing, and OpenAI is handing out access through a tier it calls Daybreak Red. The company pitches it at experienced defenders working complex authorized engagements.
OpenAI says the model has already been used in live research and turned up previously unknown vulnerabilities in widely used open source software, including Chrome’s V8 engine.
A second tier, Daybreak Blue, opens up frontier models including GPT-5.6 Sol with guardrails for a broader set of defensive work: vulnerability discovery, code review, malware analysis, incident response, and patch validation. Approved Daybreak partners can build managed services on those models for their own customers.
Up next
Claude pushed a Riemann zeta bound from 41.6% to 67.2%
An unreleased research version of Claude raised the proven lower bound on Riemann zeta zeros lying on the critical line from 41.6% to 67.2%, Anthropic says. It did not prove the hypothesis. The run took ~60 subagents, 2,400 shell commands and 31M output tokens.
Related stories
- ExploitGym score jumps to 48.5% in OpenAI’s GPT-6 Cyber
- Cyber access gate keyed to a 1996 US export list
- OpenAI’s Daybreak Red and Blue land in Amazon Bedrock
- OpenAI shelves its next model after internal safety tests
- Zero-day discovery comes with a monitoring gap
- Astra preview puts OpenAI’s safeguards to the test
Comments
No comments yet. Be the first.
Join the conversation
Sign in to leave a comment. It appears after moderation.