o1

OpenAI presents evidence that reasoning models like o1 become more robust to adversarial attacks when given more inference-time compute to think longer. The research demonstrates that increased computation reduces attack success rates across multiple task types including mathematics, factuality, and adversarial images, though significant exceptions remain.

OpenAI releases o1 model to API with production-ready features including function calling, structured outputs, vision capabilities, and 60% lower latency than o1-preview. Additional developer tools include Realtime API improvements, Preference Fine-Tuning, and new Go and Java SDKs.

OpenAI releases the o1 System Card detailing safety evaluations and preparedness framework assessments for the o1 and o1-mini models, which use chain-of-thought reasoning trained with large-scale reinforcement learning to improve safety and robustness.

OpenAI published acknowledgements for external testers and red teamers who contributed to the evaluation and safety testing of the o1 model. The document lists individuals and organizations involved in red teaming and preparedness collaboration efforts.

OpenAI released the o1 model series, designed with extended reasoning capabilities to tackle complex problems in science, coding, and math by spending more time thinking before responding.