SuperAPI AI Directory
Partner with us中文AI newsFavorites
DirectoryNews and tutorials

GPT-5.2 - OpenAI's latest general AI model series

Author: AI小集 · Original source: AI工具集

What is GPT-5.2

GPT-5.2 is OpenAI's latest advanced artificial intelligence model series, designed specifically for professional work and complex tasks. The model offers three versions: GPT-5.2 Instant is suitable for quickly handling everyday tasks such as information lookup and translation, GPT-5.2 Thinking is optimized for complex tasks such as programming, long document processing, and multi-step project management, and GPT-5.2 Pro focuses on highly difficult tasks, providing high precision and high reliability, making it suitable for scientific research and complex analysis. GPT-5.2 has set new industry standards in multiple benchmarks, such as GDPval and SWE-Bench Pro, while reducing hallucination rates and improving safety and reliability.

GPT-5.2

Main features of GPT-5.2

  • Professional work optimization: Excels in creating spreadsheets, building presentations, writing code, processing long documents, and executing complex multi-step projects, significantly improving work efficiency.
  • Powerful programming capabilities: Sets records in benchmarks such as SWE-Bench Pro, can reliably debug code, implement feature requirements, refactor large codebases, and reduce manual intervention.
  • Long text processing: Maintains coherence and accuracy in long document analysis (such as reports, contracts, and academic papers), suitable for in-depth analysis and multi-source workflows.
  • Visual understanding: Excels in chart reasoning and software interface understanding, can accurately interpret data dashboards, technical charts, and visual reports.
  • Tool calling: Reliably uses tools in long-chain, multi-turn tasks, supports end-to-end workflows, and reduces breaks between steps.
  • Science and mathematics assistance: Performs excellently in benchmarks such as GPQA Diamond and FrontierMath, can assist scientific research and solve complex mathematical problems.
  • Safety improvements: More reliable when handling sensitive topics, reduces harmful responses, and protects user safety.

Performance of GPT-5.2

  • Professional knowledge work：
    • In the GDPval benchmark, GPT-5.2 Thinking achieved a 70.9% win rate, surpassing human expert level for the first time. It performs excellently in knowledge work scenarios across 44 occupations, such as creating presentations, spreadsheets, and accounting reports.
    • In internal junior investment banking analyst tasks, the average score of GPT-5.2 Thinking was higher than GPT-5.1 by 9.3%, reaching 68.4%.
  • Programming capabilities：
    • In the SWE-Bench Pro test, GPT-5.2 Thinking achieved 55.6%, setting a new high and significantly outperforming previous versions.
    • In the SWE-bench Verified test, GPT-5.2 Thinking achieved a high score of 80%, demonstrating outstanding performance.
  • Long text processing: In the OpenAI MRCRv2 test, GPT-5.2 Thinking achieved near 100% accuracy when processing long documents up to 256k tokens, suitable for in-depth document analysis.
  • Visual understanding: In the CharXiv Reasoning and ScreenSpot-Pro tests, GPT-5.2 Thinking greatly reduced its error rate, and can more accurately interpret charts, interfaces, and visual reports.
  • Tool calling: In the Tau2-bench Telecom test, GPT-5.2 Thinking achieved 98.7% accuracy, performing excellently, and even at the lowest reasoning intensity, its performance was significantly better than GPT-5.1 and GPT-4.1.
  • Science and mathematics：
    • On the GPQA Diamond test, GPT-5.2 Pro achieved 93.2%, and GPT-5.2 Thinking achieved 92.4%, demonstrating outstanding performance.
    • On the FrontierMath test, GPT-5.2 Thinking solved 40.3% of expert-level math problems.
  • Factual accuracy: In anonymized queries on ChatGPT, GPT-5.2 Thinking reduced the error rate by 30% compared with GPT-5.1 Thinking, significantly improving factual accuracy.
  • Safety: When handling sensitive topics such as suicide and mental health, GPT-5.2 showed a significantly lower rate of undesirable responses, performing more safely and reliably.
GPT-5.2

How to use GPT-5.2

  • In ChatGPT use GPT-5.2: GPT-5.2 has begun rolling out today to paid users (including the Plus, Pro, Go, Business, and Enterprise plans), covering the three versions Instant, Thinking, and Pro. To ensure smoothness and stability, the deployment will be carried out in phases, and some users may not be able to use it immediately for the time being. During this period, paid users can still continue to use GPT-5.1 (legacy mode) for three months, after which GPT-5.1 will be gradually retired. Users can choose the version that suits their needs.
  • Using GPT-5.2 on the API platform: On the OpenAI API platform, all three versions of GPT-5.2 are fully available, with the specific names: GPT-5.2 Instant (gpt-5.2-chat-latest）、GPT-5.2 Thinking（gpt-5.2) and GPT-5.2 Pro (gpt-5.2-pro). Developers can set reasoning parameters in GPT-5.2 Pro. Both GPT-5.2 Pro and GPT-5.2 Thinking support the new fifth-level reasoning effort xhigh, suitable for tasks with extremely high quality requirements. Through the API, users can flexibly integrate the powerful capabilities of GPT-5.2 to meet the needs of complex application scenarios.

GPT-5.2 project address

  • Project official website：https://openai.com/index/introducing-gpt-5-2/
  • Technical paper：https://cdn.openai.com/pdf/3a4153c8-c748-4b71-8e31-aecbde944f8d/oai_5_2_system-card.pdf

GPT-5.2 product pricing

ChatGPT subscription prices remain unchanged

API pricing

  • GPT-5.2 Instant（gpt-5.2-chat-latest）：
    • Input: $1.75/million tokens
    • Output: $14/million tokens
  • GPT-5.2 Thinking（gpt-5.2）：
    • Input: $1.75/million tokens
    • Output: $14/million tokens
  • GPT-5.2 Pro（gpt-5.2-pro）：
    • Input: $21/million tokens
    • Output: $168/million tokens
  • For cached inputs, GPT-5.2 offers a 90% discount, namely: cached input: $0.175/million tokens

GPT-5.2 application scenarios

  • Professional document processing: GPT-5.2 can efficiently process long documents, such as reports, contracts, and academic papers, support in-depth analysis and cross-document information integration, maintain content coherence and accuracy, and significantly improve work efficiency.
  • Programming and software development: Generate, debug, and optimize code in multiple programming languages, support complex front-end development and large-scale codebase refactoring, provide strong assistance to full-stack engineers, and accelerate the software development process.
  • Scientific research and data analysis: Assist in solving advanced mathematical problems and answering complex scientific questions, support in-depth data analysis and report generation, provide scientific researchers with a powerful tool, and advance academic research.
  • Customer service and support: Through multi-turn conversations and tool calls, handle complex customer issues, such as flight delays and lost luggage, provide end-to-end solutions, and improve customer satisfaction.
  • Content creation and copywriting: Quickly generate high-quality copywriting, news reports, social media content, and creative writing, supporting multiple languages and styles, inspiring creative inspiration, and meeting diverse content needs.