OpenAI Upgrades ChatGPT Research with GPT-5.2
OpenAI has upgraded ChatGPT's Deep Research feature to run on its GPT-5.2 model, enhancing reasoning and providing users with expanded controls. The update enables users to fine-tune research agent behavior and review verifiable sources. The move comes as OpenAI leadership signals a shift toward sustainable monetization, including potential advertising, and faces scrutiny over the safety of its upcoming GPT-5.3-Codex model in California.
- The Deep Research feature was initially powered by OpenAI's o3 model, later incorporating a more cost-efficient o4-mini variant before the upgrade to GPT-5.2. A key improvement with the latest update is the ability for users to track research progress in real-time and interrupt the process to refine it with new prompts or sources. - GPT-5.2, released on December 11, 2025, is a family of models optimized for professional work, including "Instant" for speed, and "Thinking" and "Pro" for advanced reasoning. It demonstrates state-of-the-art performance in coding, vision, and understanding long documents. - The scrutiny over GPT-5.3-Codex in California comes from allegations by a watchdog group, The Midas Project, which claims OpenAI failed to implement required safeguards for a model classified as a high cybersecurity risk. OpenAI has disputed the allegations, stating the model completed its full governance process and did not demonstrate long-range autonomy capabilities. - OpenAI's new CEO of Applications, Fidji Simo, previously the CEO of Instacart and a former executive at Meta, is leading the company's monetization efforts. Her team is exploring various strategies, including a recent job posting for a "Growth Paid Marketing Platform Engineer" to build an in-house system for ad integration. - The user interface for Deep Research has been updated with a fullscreen document viewer that includes a table of contents and a column for citations. This allows users to review and export the final reports in formats like Markdown, Word, and PDF. - In late 2024, OpenAI experienced significant leadership changes, including the departure of Chief Technology Officer Mira Murati and other research executives. These changes have led analysts to suggest a strategic shift from a "research-first" to a "product-first" organization focused on commercialization. - Competitors are actively developing similar deep research functions. Alternatives include Google's Gemini Deep Research, Anthropic's Claude, and Perplexity AI, which are often compared based on speed, depth of analysis, and integration with other ecosystems.
