_February 10, 2026 update:_ _You can now connect deep research to any MCP or app and restrict web searches to trusted sites, so you can focus on authenticated, industry-standard sources. You can also now track progress in real-time and interrupt to refine with follow-up prompts or new sources. We've updated the visual experience so it's easier to start, track, and review your research from end to end._ _July 17, 2025 update:_ _Deep research can now go even deeper and broader with access to a visual browser as part of ChatGPT agent. To access these updated capabilities, simply select “agent mode” from the dropdown in the composer and enter your query directly. The original deep research functionality remains available via the “deep research” option in the tools menu._ _April 24, 2025 update__: We’re significantly increasing how often you can use deep research—Plus, Team, Enterprise, and Edu users now get 25 queries per month, Pro users get 250, and Free users get 5. This is made possible through a new lightweight version of deep research powered by a version of o4-mini, designed to be more cost-efficient while preserving high quality. Once you reach your limit for the full version,
核心贡献 · Key contributions
在 ChatGPT 中推出深度研究智能体，可进行多步骤网络研究，综合数百个来源生成分析师级报告。 Launches deep research agent in ChatGPT for multi-step web research, synthesizing hundreds of sources into analyst-level reports.
由针对浏览和数据分析优化的 o3 版本驱动，使用强化学习来规划并执行研究轨迹。 Powered by a version of o3 optimized for browsing and data analysis, using reinforcement learning to plan and execute research trajectories.
在 GAIA 基准测试上达到最先进水平，在人类最后考试中取得 26.6%的准确率，在化学、人文学科和数学上优于先前模型。 Achieves state-of-the-art on GAIA benchmark and 26.6% on Humanity's Last Exam, outperforming prior models in chemistry, humanities, and math.
支持文件上传、Python 工具使用、嵌入图像/图表，以及实时进度跟踪，可中断并优化。 Supports file uploads, Python tool use, embedded images/graphs, and real-time progress tracking with interrupt and refine capabilities.
展示了使用双模式吸附模型对玻璃态聚合物进行竞争性吸附预测，包含纯气体和混合气体的方程。 Demonstrates competitive sorption prediction using dual-mode sorption model for glassy polymers, with equations for pure and mixed gases.
为 iOS 翻译应用提供详细市场分析，推荐日本、美国、英国、加拿大、澳大利亚作为首要目标市场。 Provides detailed market analysis for iOS translation app, recommending Japan, US, UK, Canada, Australia as primary targets.
局限 · Limitations
可能产生事实幻觉或错误推断，但内部评估显示比率低于现有 ChatGPT 模型。 May hallucinate facts or make incorrect inferences, though at lower rate than existing ChatGPT models per internal evaluations.
难以区分权威信息与谣言，且在置信度校准方面表现薄弱。 Struggles with distinguishing authoritative information from rumors and shows weakness in confidence calibration.
目前计算密集；仅 Pro 用户享有高使用限制（每月 100 次查询），其他层级推出较慢。 Currently compute-intensive; high usage limits only for Pro users (100 queries/month), with slower rollout to other tiers.
双模式吸附模型参数非通用；多个拟合可能产生不同的混合气体预测，限制可靠性。 Dual-mode sorption model parameters are not universal; multiple fits can yield different mixed-gas predictions, limiting reliability.
模型忽略聚合物溶胀/塑化，无法捕捉 S 形等温线或协同吸附效应。 Model ignores polymer swelling/plasticization and cannot capture sigmoidal isotherms or synergistic sorption effects.