OpenAI Launches GPT-5.2, Its Most Capable Model Yet for Knowledge Work

OpenAI has released GPT-5.2, calling it its most capable model series yet for professional knowledge work. The launch, on December 11, 2025, came amid intensifying competition at the frontier of AI.

Three modes plus a coding variant

GPT-5.2 arrives in three modes: GPT-5.2 instant, GPT-5.2 thinking (with standard and extended reasoning) and GPT-5.2 Pro, the latter two being reasoning models. OpenAI also shipped GPT-5.2-Codex, an agentic coding model tuned for long-horizon work through context compaction, stronger performance on large refactors and migrations, improved behaviour in Windows environments, and enhanced cybersecurity capabilities.

Benchmarks and the ‘Code Red’ backdrop

OpenAI said GPT-5.2 was designed to be stronger at building spreadsheets and presentations, writing code, perceiving images, handling long context and using tools across complex, multi-step projects. The company reported state-of-the-art results across several benchmarks, including GDPval. The release landed roughly three weeks after Google’s Gemini 3 Pro, and media reports pointed to an internal “Code Red” memo — triggered by Gemini’s strength — as a factor in accelerating the model’s launch from an originally planned late-December window.