OpenAI has introduced GPT‑5.2, the newest upgrade to its GPT‑5 model family, making it the most capable series yet for professional knowledge work, long-context understanding, coding, vision, and tool use. The model delivers significant performance improvements across complex tasks and sets new benchmarks for AI utility in real-world workflows.
According to OpenAI, GPT‑5.2 Thinking, the variant optimised for deeper reasoning, outperforms previous models on multiple professional benchmarks. On GDPval, a professional knowledge work benchmark spanning 44 occupations, GPT‑5.2 Thinking ties or surpasses top industry performance. It also achieves 100 per cent on the AIME 2025 benchmark without tools and shows gains in scientific figure reasoning and abstract problem-solving compared with earlier GPT‑5 models.
GPT‑5.2 is available to paid ChatGPT subscribers via Instant, Thinking, and Pro tiers and through the API for developers. Earlier versions such as GPT‑5.1 remain accessible during the transition period. OpenAI emphasises that this release is part of its ongoing roadmap for improving AI reliability and professional usability.
Some media reports note that GPT‑5.2 deployment comes amid heightened competition in generative AI, particularly from Google’s Gemini 3. Certain outlets highlighted an internal “red code” alert at OpenAI in early December, signalling accelerated timelines for testing and rollout, although OpenAI frames the update as a planned iteration in its model series.
The model’s long-context reasoning allows coherent and accurate output across extended documents, research reports, and multi-file projects. GPT‑5.2’s vision and tool-calling capabilities have also been strengthened, supporting tasks involving images, code, and automated workflows. Observers note that these improvements position GPT‑5.2 as a more reliable partner for knowledge workers, developers, and enterprise teams seeking advanced analytical support.
