modelsNov 19
GPT-5.1-Codex-Max System Card
OpenAI published a system card for GPT-5.1-Codex-Max, detailing safety measures for the model. The card outlines model-level and product-level mitigations to prevent misuse. Builders integrating GPT-5.1-Codex-Max should review these measures to ensure safe deployment. The system card is a technical document aimed at transparency.
Key takeaways
- GPT-5.1-Codex-Max has specialized safety training for harmful tasks.
- The model features agent sandboxing and configurable network access.
- OpenAI aims for transparency with detailed system documentation.