OpenAI announced on August 10, 2026 that its Daybreak program will now offer a general‑purpose ‘Blue’ tier and a specialized ‘Red’ tier, along with a new GPT‑5.6‑Cyber model designed to reduce refusals on advanced security tasks.

OpenAI on August 10, 2026 unveiled a two‑tier expansion of its Daybreak program, adding a general‑purpose “Blue” tier, a specialized “Red” tier, and a new GPT‑5.6‑Cyber model aimed at handling advanced security tasks with fewer refusals.

What the New Daybreak Tiers Offer

The Blue tier is positioned as a versatile, general‑purpose offering that retains the core capabilities of the existing Daybreak service while providing broader access to developers and enterprises.

The Red tier, by contrast, is tailored for high‑risk environments such as penetration testing, threat hunting, and incident response, delivering deeper access to specialized prompts and model configurations.

Introducing GPT‑5.6‑Cyber

Alongside the tiered rollout, OpenAI launched GPT‑5.6‑Cyber, a model engineered to reduce the frequency of refusals when processing complex cybersecurity queries, while still respecting safety guardrails.

The model incorporates an updated refusal‑reduction algorithm that distinguishes between benign investigative prompts and genuinely malicious intent, aiming to support security professionals without compromising ethical standards.

Key Differences Between Blue and Red

  • Access level: Blue provides standard API quotas; Red includes higher rate limits and priority support.
  • Prompt library: Red unlocks a curated set of security‑focused prompts and response templates.
  • Compliance: Red tier customers receive dedicated compliance documentation for regulated industries.

Implications for the Cybersecurity Landscape

By offering a dedicated tier for advanced security work, OpenAI aims to become a more integral part of the cyber‑defense stack, potentially reducing reliance on legacy, on‑premise tools.

Analysts note that the refusal‑reduction focus of GPT‑5.6‑Cyber could streamline automated threat analysis workflows, though they also caution about the need for robust monitoring to prevent misuse.

OpenAI’s expansion reflects a broader industry trend toward AI‑augmented security operations, positioning the company as a key player as the cyber‑defense window narrows.

OpenAI announcement on expanding Daybreak and the new cyber model