GPT-5.2: Advanced Language Model by OpenAI
GPT-5.2 has been officially launched by OpenAI, being the most sophisticated model for professional knowledge work and advanced reasoning available from the company. This release on 11th December comes right after increasing competition from Google, whose Gemini 3 model has led OpenAI’s CEO, Sam Altman, to declare an internal “code red.”
Three Tiers of GPT-5.2
The GPT-5.2 model is available in three distinct tiers: Instant, Thinking, and Pro. These tiers are specifically designed to accommodate a wide range of computational requirements, from quickly retrieving information to managing intricate, multi-step projects. OpenAI asserts that this model includes considerable performance improvements in vital areas essential for enterprise applications.
Exceptional Mathematical Skills
The model showcases groundbreaking mathematical abilities. OpenAI revealed that the mid-range Thinking version attained an unprecedented score on the FrontierMath benchmark, accurately solving 40.3% of college-level problems. In a notable case study, the top-tier GPT-5.2 Pro successfully solved a simplified version of an open question in statistical learning theory, with human experts focused exclusively on verification.
Enhanced Software Engineering Capability
Besides its intellectual prowess, GPT-5.2 has significantly improved performance in software engineering. It achieved a new benchmark score on the challenging SWE-Bench Pro coding evaluation. The model also shows marked advancements in interpreting charts and figures from scientific articles, boosting accuracy on the CharXiv Reasoning benchmark by over 8% compared to its predecessor.
Positioning as a Work Partner
OpenAI is promoting GPT-5.2 as a dependable collaborator. The Thinking version reportedly secured a win rate of over 70% when evaluated against top industry experts on well-defined knowledge work tasks, including spreadsheet creation and presentation development. Now accessible to ChatGPT subscribers and via the developer Application Programming Interface (API), the launch underscores the company’s commitment to regaining traction in the dynamic AI landscape.