By skyforbes summary:
OpenAI’s GPT-5.2 is a new frontier model (Instant, Thinking, Pro) focused on professional, long-running, tool-using workflows, with strong gains in reasoning, coding, long-context, and vision. It outperforms GPT-5.1 and many human experts on benchmarks like GDPval (knowledge work), SWE-Bench (coding), MRCRv2 (long documents), GPQA and FrontierMath (science/math), and ARC-AGI-2 (abstract reasoning), while hallucinating less. In practice, it more reliably produces high-quality spreadsheets, presentations, code, and end-to-end agentic workflows, and better understands charts, interfaces, and technical diagrams. GPT-5.2 also adds stronger safety behavior—especially for mental health and minors—and is now rolling out to paid ChatGPT plans and available via API (gpt-5.2, gpt-5.2-chat-latest, gpt-5.2-pro) at a higher per-token price than GPT-5.1 but with better overall efficiency.