OpenAI Reveals How Much Its Researchers Are Spending on AI Coding

OpenAI has released a detailed update regarding the internal use of AI agents to accelerate its own research and development processes. According to a report from OpenAI titled "Research acceleration: The view inside OpenAI," the company has reached a goal originally set by Sam Altman in 2025 to create an "automated research intern." This internal tool, described by PCMag as a "Research Assistant," utilizes AI to build better AI, essentially creating a loop where the company's own technology is used to speed up the development of future iterations. This effort is focused on integrating AI agents directly into the research workflow to optimize how the organization operates. Coverage from Business Insider and analyticsindiamag.com highlights the significant financial investment associated with this acceleration, reporting that OpenAI's top researchers are spending more than $7,000 per day on Codex. This specific data point emphasizes the high cost of the compute and API usage required for high-level AI coding.

While YourStory.com reports that these AI agents are helping to increase the speed of research, Fortune notes a tension within the company, stating that this acceleration is happening even as the company's chief scientist has called for a slowdown in progress. To understand why this development is significant, it is necessary to look at the timeline established in the reporting. The Times of India notes that the objective for the "automated research intern" was set a year prior, in 2025, marking the current update as a milestone in the company's internal operational strategy. This reflects a broader trend of AI companies using their own generative tools to automate technical labor. However, The New Stack provides a contrasting perspective on the outcome of this automation, suggesting that the deployment of AI agents is actually creating more work rather than less for the human staff involved.

Future developments will likely center on whether the internal efficiency gains justify the high daily expenditures on Codex. Based on the current coverage, observers should monitor how OpenAI balances the drive for acceleration with the concerns raised by its chief scientist. The company has now published a long update on these efforts, which provides a view into the "innermost loop" of its research cycle. As reported by Engadget, the formal goal of creating an automated research intern has been met, leaving the company to manage the resulting workload and financial costs associated with these high-spending research habits.