OpenAI officially released a new generation of cutting-edge model series GPT-5.2, positioning itself as the most powerful flagship model to date, mainly for professional knowledge work and long-term operating agent scenarios. Officially, GPT-5.2 has significant improvements over its predecessors in creating spreadsheets, presentations, code writing, image understanding, and complex multi-step task processing.
In a number of authoritative benchmarks, GPT-5.2 Thinking reached or exceeded the level of human experts in GDPval evaluations covering 44 occupations, and set new model records in tests such as software engineering, scientific Q&A, advanced mathematics, abstract reasoning, and long document comprehension.
In terms of availability, GPT-5.2 Instant, Thinking, and Pro will be available to Plus, Pro, Go, Business, and Enterprise paid users in ChatGPT, and GPT-5.1 will be retained in ChatGPT for about three months before discontinuing support. The corresponding names on the API side are gpt-5.2-chat-latest, gpt-5.2 and gpt-5.2-pro, and the price of the basic version of GPT-5.2 is $1.75 per million input tokens and $14 per million output tokens.
Frequently Asked Questions
Q: What problems does OpenAI's GPT-5.2 mainly solve?
A: GPT-5.2 is mainly aimed at professional knowledge-based work and complex multi-step tasks, improving document processing, data analysis, presentation design, coding, and long-context reasoning capabilities.
Q: What are the core improvements of GPT-5.2 compared to GPT-5.1 in terms of performance?
A: GPT-5.2 achieves higher results on benchmarks such as GDPval, SWE-bench, GPQA, FrontierMath, and ARC-AGI, with better performance in long document understanding and factual performance, and significantly lower hallucination rates.
Q: How can users use GPT-5.2 series models in ChatGPT?
A: ChatGPT paid users can choose ChatGPT-5.2 Instant, Thinking, or Pro in the model selection, and the function will be available to Plus, Pro, Go, Business, and Enterprise in a gradual rollout.
Q: What is the naming and billing method of GPT-5.2 in the API?
A: GPT-5.2 Thinking corresponds to gpt-5.2, Instant corresponds to gpt-5.2-chat-latest, and Pro corresponds to gpt-5.2-pro, with the basic version being billed at $1.75 per million inputs and $14 per output, with significant discounts on cached inputs.
Q: How will the status of GPT-5.1 change in the product line after this release?
A: GPT-5.1 will continue to be available to paying users in the form of a traditional model in ChatGPT for about three months, after which support will be discontinued in ChatGPT, but no retirement timeline has been announced in the API.
