OpenAI Details GPT-5.5 Instant Safety

OpenAI has published the GPT-5.5 Instant System Card, outlining the safety framework for its newest rapid-response AI model. This release details the approach to mitigate risks associated with the advanced capabilities of GPT-5.5 Instant.

The safety protocols for GPT-5.5 Instant build upon previous models in the series. However, this is the first Instant model to be categorized as 'High capability' in critical areas like Cybersecurity and Biological & Chemical Preparedness.

Heightened Safety Classifications

This classification necessitates more stringent safeguards. OpenAI refers to GPT-5.5 Instant as gpt-5.5-instant in the documentation.

The primary model for comparison is GPT-5.3 Instant, with no GPT-5.4 Instant model released.

For clarity, the core GPT-5.5 model is designated as GPT-5.5 Thinking.