GPT-5.2

GPT-5.2, released in December 2025, represents OpenAI's continued refinement of the GPT-5 architecture with dramatic improvements in reliability and context handling. The model features a 256K token context window with near-perfect accuracy retention, expandable to 400K tokens maximum, enabling it to process extremely long documents without degradation.

The model achieves perfect 100% accuracy on AIME 2025, a remarkable milestone for mathematical reasoning, and reduces its hallucination rate to just 6.2%-a significant improvement in factual reliability. On GDPval (a benchmark comparing AI to human experts), GPT-5.2 beats or ties human experts 70.9% of the time, demonstrating superhuman performance in specific domains.

GPT-5.2 is available in specialized variants: GPT-5.2 Instant (optimized for low latency) and GPT-5.2 Thinking (with extended reasoning capabilities). These variants allow users to choose the appropriate trade-off between response speed and reasoning depth, making the model adaptable to different application requirements from real-time chat to complex analytical tasks.