OpenAI Announces GPT-5.2

Image on Unsplash

OpenAI announced GPT-5.2 on Thursday, marking an accelerated response to Google's Gemini 3. Gemini displaced OpenAI from the top position on leaderboards in November and prompted CEO Sam Altman to declare code red at the company..

The new model comes in three variants. ‘Instant’ for fast information retrieval and writing, ‘Thinking’ for complex tasks like coding and planning, and ‘Pro’ for highest-accuracy scenarios requiring advanced reasoning.

OpenAI claims that GPT-5.2 outperforms competitors on several metrics. It scored 55.6% on SWE-Bench Pro, which measures software engineering capability, compared to Gemini 3 Pro's 43.3%. It’s a 12.3% advantage in the most visible benchmark between the two systems.

The release comes less than two months after Gemini 3's launch, which is an unusually rapid development and launch cycle. It suggests that OpenAI has accelerated its internal timelines in response to competitive pressure.

OpenAI has positioned GPT-5.2 as the smartest generally-available model in the world, though this claim is difficult to evaluate because different benchmarks measure different capabilities and companies emphasize benchmarks where their models perform best.