CATCH #609
DETECTED AUG 10, 2026 17:39 UTC
THE CATCHscore 10/10
EXHIBIT 609 · FLASH
Aug 10, 2026 17:39 UTC
openapi+pages
https://raw.githubusercontent.com/openai/openai-openapi/master/openapi.yaml
New OpenAI models 'gpt-5.6-cyber' and 'Daybreak Blue' spotted in API spec and changelog, focused on cybersecurity tasks and defensive work, requiring special approval.
READY TO POST:
looks like openai is quietly rolling out new models, gpt-5.6-cyber and daybreak blue. very cool to see specific models for cybersecurity tasks.
openai's new daybreak program seems to be for specialized cyber models like gpt-5.6-cyber. definitely one to watch if you're in infosec.
EVIDENCE — the receipts
[MERGED across 2 sources: openapi, pages]
shared signal: gpt-5.6, gpt-5.6-cyber
--- openapi :: https://raw.githubusercontent.com/openai/openai-openapi/master/openapi.yaml ---
OpenAI API spec (https://raw.githubusercontent.com/openai/openai-openapi/master/openapi.yaml): 1 additions to the official API spec
[new models] (1)
+ gpt-5.6-cyber
--- pages :: https://platform.openai.com/docs/changelog ---
OpenAI changelog changed: https://platform.openai.com/docs/changelog
18 added lines
---
+++
@@ -255,2 +255,3 @@
Private Link
+IP allowlist
Workload identity federation
@@ -702,2 +703,19 @@
August, 2026
+Aug 7
+Feature
+gpt-5.6-cyber
+daybreak-red-latest
+daybreak-blue-latest
+v1/responses
+Released
+GPT-5.6 Cyber
+, available through
+Daybreak Red
+. Built on GPT-5.6 Sol, it is trained to improve capabilities on several specialized cybersecurity tasks (for example, finding zero-day exploit chains) and to reduce refusals for highly dual-use cyber tasks.
+Released
+Daybreak Blue
+, an alias for our frontier general-purpose models, with safeguards calibrated for defensive cybersecurity work.
+These models require separate approval and provisioning. You can apply to join the Daybreak program
+here
+.
Aug 6
caught by the 20-minute sweep ·
the live wire · all receipts