OpenAI has launched ChatGPT-5.2, a new family of AI models designed to enhance reasoning and productivity, particularly for professional tasks. The release follows an internal alert from CEO Sam Altman about competition from Google's Gemini 3. The update includes three variants aimed at different user needs, starting with paid subscribers.
OpenAI announced the release of GPT-5.2 on Thursday, December 11, 2025, introducing updates focused on improving AI capabilities for work-related activities. The company positions the model as its strongest yet for science and math, with enhancements in creating spreadsheets, building presentations, writing code, perceiving images, understanding long contexts, using tools, and handling complex, multi-step projects.
This launch comes after CEO Sam Altman issued an internal 'code red' memo last week, highlighting the threat from Google's Gemini 3 chatbot, released in November. The memo urged employees to prioritize improvements to ChatGPT amid reversing competitive dynamics, echoing Google's past response to ChatGPT's debut.
GPT-5.2 is available in three tiers: GPT-5.2 Instant for fast, everyday tasks like writing and translation; GPT-5.2 Thinking for advanced professional work such as coding and math; and GPT-5.2 Pro for high-accuracy answers to difficult questions. It features a 400,000-token context window, enabling processing of hundreds of documents, and a knowledge cutoff of August 31, 2025.
'We designed GPT-5.2 to unlock even more economic value for people,' said Fidji Simo, OpenAI's chief product officer, during a press briefing. The model also supports developers in building AI agents through better general intelligence, long-context understanding, tool-calling, and vision.
Benchmarks show GPT-5.2 Thinking scoring 55.6 percent on SWE-Bench Pro, surpassing Gemini 3 Pro's 43.3 percent, and 92.4 percent on GPQA Diamond versus 91.9 percent for Gemini 3 Pro. On the GDPval benchmark, it matches or exceeds human professionals on 70.9 percent of tasks across 44 occupations, at over 11 times the speed and under 1 percent of the cost. It generates 38 percent fewer confabulations than GPT-5.1.
The rollout begins with paid ChatGPT plans and API access for developers, priced at $1.75 per million input tokens—a 40 percent increase from GPT-5.1, which remains available for three months.
