OpenAI has significantly broadened access to its powerful AI research assistant, making advanced information gathering capabilities available to a much wider audience of subscribers. Initially restricted to the highest-tier subscribers, this expansion represents a strategic move to bring sophisticated AI research tools to more users while maintaining tiered access privileges.
Wider Access for Paid Subscribers
OpenAI has officially extended its Deep Research feature to all paying ChatGPT subscribers, including Plus, Team, Edu, and Enterprise users. Initially launched earlier this month exclusively for Pro tier subscribers at US$200 per month, the feature is now accessible to users on more affordable subscription plans. While Pro subscribers retain the highest usage allocation with 120 queries per month (up from the original 100), other paying users will receive a more limited allowance of 10 deep research queries monthly. This tiered approach maintains exclusivity for the premium tier while expanding accessibility.
Deep Research Access By Subscription Tier
- ChatGPT Pro: 120 queries per month (US$200/month)
- ChatGPT Plus: 10 queries per month
- ChatGPT Team: 10 queries per month
- ChatGPT Edu: 10 queries per month
- ChatGPT Enterprise: 10 queries per month
Powerful Research Capabilities
Deep Research functions as an autonomous AI research agent capable of conducting multi-step research by gathering comprehensive information from across the web. Powered by a specialized version of OpenAI's o3 model optimized for web browsing and data analysis, the tool can search and interpret massive amounts of online content, including texts and images, synthesizing findings into comprehensive reports. The system works independently once prompted, effectively serving as a personal research analyst that can complete in minutes what would typically take humans hours to accomplish.
Enhanced Features and Performance
Along with expanding access, OpenAI has introduced several improvements to Deep Research. The tool now includes embedded images with citations in its outputs and demonstrates better capabilities for understanding and referencing uploaded files. Each report includes clear citations and a summary of the agent's reasoning process, allowing users to verify information independently. This transparency is particularly important given that OpenAI acknowledges the system can occasionally hallucinate facts or make incorrect inferences, though reportedly at lower rates than existing ChatGPT models.
Impressive Benchmark Results
Deep Research has demonstrated impressive performance in standardized AI evaluations. The system outperformed several advanced AI models, including GPT-4o, Grok-2, Claude 3.5 Sonnet, and Gemini Thinking on Humanity's Last Exam, a benchmark created by Scale AI and the Center for AI Safety. With a 26.6% accuracy score, Deep Research even surpassed o3-mini high, which had previously held the highest score. Additionally, the tool topped the GAIA external leaderboard, a public benchmark that evaluates AI on real-world questions.
Market Competition
The research agent space is becoming increasingly competitive. Google offers a similar feature (also called Deep Research) through its Gemini Advanced service for US$20 per month via the Google One AI Premium plan. Elon Musk's xAI has launched DeepSearch, available to X Premium (US$8 monthly/US$84 yearly) and Premium+ (US$40 monthly/US$395 yearly) subscribers. Meanwhile, Microsoft has introduced Think Deeper, which leverages OpenAI's O1 reasoning model but lacks internet access and autonomous capabilities. Unlike its competitors, Microsoft's offering is available for free.
Competitive Research Agent Pricing
- OpenAI Deep Research (Pro): US$200/month
- Google Deep Research (Gemini Advanced): US$20/month
- xAI DeepSearch (X Premium): US$8/month or US$84/year
- xAI DeepSearch (X Premium+): US$40/month or US$395/year
- Microsoft Think Deeper: Free
Target Users and Limitations
OpenAI positions Deep Research as particularly valuable for professionals engaged in intensive knowledge work across finance, science, policy, and engineering sectors. The tool excels at finding niche information that would typically require multiple manual searches. However, OpenAI cautions that the system may struggle to distinguish authoritative information from rumors and can sometimes fail to properly convey uncertainty, emphasizing the continued importance of human oversight and verification.