Overview
- GPT-5.2 begins rolling out today to ChatGPT Plus, Pro, Go, Business and Enterprise users with a gradual deployment for reliability, and it is also available through the API.
- The series adds three targeted options: Instant for faster writing and information-finding, Thinking for coding, math, planning and long documents, and Pro for maximum accuracy on hard questions.
- OpenAI cites top scores on benchmarks such as SWE-Bench Pro, GPQA Diamond and its GDPval test, with GPT-5.2 Thinking reducing hallucinations by 38% versus GPT-5.1 on factual assessments.
- The launch follows OpenAI’s internal code red to prioritize ChatGPT under competitive pressure from Google’s Gemini 3, and Sam Altman told CNBC he expects the company to exit code red by January.
- OpenAI is starting an early rollout of an age-prediction system to better protect teens, with an adult mode targeted for the first quarter of 2026, and pre-release testers included Notion, Box, Shopify, Harvey, Zoom and Databricks.