OpenAI Quietly Releases GPT-5.2: New Flagship Model Achieves 71% Success Rate on Complex Tasks

15.01.2026

OpenAI has discreetly launched a new flagship model designed for advanced intellectual tasks — GPT-5.2. This release marks a significant leap in AI capabilities, with substantial improvements across multiple performance metrics.

Key Features and Improvements:

• Enhanced Problem-Solving Performance: The new model surpasses human performance in 71% of complex tasks, compared to the previous version which barely reached 39% success rate.

• Advanced Coding Capabilities: GPT-5.2 demonstrates significantly improved programming abilities, capable of generating sophisticated frontend interfaces for websites and applications with ease.

• Expanded Context Window: The model features a substantially increased context length, enabling it to better comprehend conversation flow and maintain coherence even after processing extensive message threads.

• Updated Knowledge Base: Training data has been refreshed through August 2025, ensuring access to current information.

• Reduced Hallucination Rate: The model exhibits improved accuracy with 30% fewer hallucinations compared to GPT-5.1, resulting in more reliable outputs.

The neural network is currently being rolled out to paid subscribers. Users can test the new model at the official ChatGPT platform.

Source: https://chatgpt.com/

Tags: ChatGPT OpenAI large language models AI development GPT-5.2

Share: VK Telegram Twitter