Summary

GPT-5.5 Instant (referred to as gpt-5.5-instant) is the new Instant-family model published in the system card on 2026-05-05. Its safety mitigation approach follows the established Instant-series practices but this is the first Instant model classified as High capability for Cybersecurity and Biological & Chemical Preparedness. The system card clarifies naming (no GPT-5.4 Instant; baseline is GPT-5.3 Instant) and distinguishes GPT-5.5 Thinking (the non-instant variant).

Key Points

• Classification: treated as High capability in Cybersecurity and Biological & Chemical Preparedness — apply stricter review and controls where applicable.
• Mitigations: uses the same comprehensive safety mitigation framework used in prior Instant models, with added safeguards for the elevated risk categories.
• Naming and baselines: model identifier is gpt-5.5-instant; compare behavior and performance to gpt-5.3-instant for baselining.
• Operational guidance: include additional pre-deployment risk assessment, enhanced monitoring, and constrained tooling for sensitive workflows.
• Deployment checklist (engineers):
  • Update risk assessments to reflect "High" category designations.
  • Reuse prior Instant mitigation patterns (filters, prompts, guardrails) and add tightened controls for cybersecurity/bio risks.
  • Log and monitor for misuse signals and escalate per incident response playbooks.
• Documentation: refer to the official system card (published 2026-05-05) for full details and follow organizational compliance reviews before production use.