GPT-5.6 Cyber — Specs, Pricing & Limits

GPT-5.6 Cyber is the purpose-trained security model behind Daybreak Red, OpenAI's separately approved access tier for authorized vulnerability reproduction, exploit validation, penetration testing and red teaming. Unlike the Cyber models before it, it has a full model page and a published Standard price row — but publishing a price isn't the same as offering the model to anyone with a key. It requires separate approval and provisioning through the Daybreak program's trusted-access process, which is why no calculator on this site offers it.

Most defensive work doesn't need it. OpenAI points approved defenders to Daybreak Blue first, which gives access to general-purpose models such as GPT-5.6 Sol for vulnerability discovery, secure code review, detection engineering, incident response and patch validation. Daybreak Red, with this model, is the narrower tier for work that needs a purpose-trained model, such as exploit validation and red teaming.

What's published is specific. Its context window is smaller than GPT-5.6 Sol's while its output ceiling and knowledge cutoff match it, and its price row includes a cache-write rate. A couple of gaps are recorded rather than filled. Its model page carries the long-context repricing sentence, but the pricing page publishes no long-context band for it, and since the two surfaces don't reconcile this site records no long-context rule. No reasoning-effort sentence appears on its page at all.

Its predecessors show how these entries move. GPT-5.5 Cyber has a price row but no model page, and GPT-5.4 Cyber was removed from the API without ever being priced — see what this site could not confirm.

The tables above show what the facts module holds for this model: the published Standard price row, its specs, and its rate-limit table.