OpenAI reports 3.1 agent-workdays per human workday
OpenAI says its research organization now runs 3.1 agent-workdays of automated effort for every human workday, and that it has hit its self-set milestone of an "automated research intern."
OpenAI published what it calls a snapshot of its progress toward recursive self-improvement, disclosing internal usage figures for coding agents inside its research organization.
The company says it has met the goal, announced last fall, of having an "automated research intern" by September 2026 — defined as a system that can carry out well-defined research tasks under human direction, including work that would take a skilled researcher a few days. It says it is making "strong progress" toward a fuller automated AI researcher by March 2028.
The numbers: at the start of the year, the median researcher ranked by agent usage used coding agents in modest amounts. By mid-August, that median researcher was spending more than $600 per day of inference at API prices, with the 90th-percentile user in the research org above $7,000 per day. Before June 2026, total agent runtime across the research organization was below total human labor; as of mid-August, OpenAI counts 3.1 agent-workdays for every human workday. Highly concurrent workflows — four or more agents at once — are increasing.
OpenAI also says that after "the recent Hugging Face incident" it paused reinforcement-learning training on its latest models intended for deployment while it hardened and red-teamed research environments and widened monitoring. Some workloads later resumed under stronger controls; others stayed paused.
The post does not detail that incident, the underlying measurement methodology beyond calling it preliminary, or which models are affected. OpenAI argues labs should be required to publicly track RSI progress, and says it will keep disclosing regardless.