OpenAI says its internal coding-agent systems have reached what it calls an “automated research intern” milestone.
The company describes agents that can complete well-defined research tasks under human direction, including work it estimates could take a skilled researcher several days. The claim refers to OpenAI’s internal research workflow, not a generally available autonomous-research product.
OpenAI also disclosed very high internal agent usage. By mid-August, median researcher usage exceeded $600 per day of inference at API prices, while the 90th percentile exceeded $7,000 per day. The company estimates 3.1 agent-workdays of runtime for every human workday in its research organization.
Those figures are not equivalent to a 3.1× productivity gain. OpenAI itself says the measurements are preliminary and that people still choose research priorities, judge results and decide when to continue, pause or deploy work.