OpenAI’s post Research acceleration: a view inside OpenAI (openai.com) reports the automated research intern delivering 3.1 agent-workdays per human workday, up from under one before June, and sets March 2028 as the target for a fully automated researcher; a separate analysis put internal time horizons up 6.18x since January, a doubling every 2.2 months. One observer ranked the shift among the biggest in the history of science, and unnoticed by most. It is the delivery on the research intern by September that OpenAI promised in March and the automated researcher by March 2028 it planned in June, and it gives a unit to the loop the previous day’s AGI ship date described, a month after the lab said its R&D evaluations had saturated.
Development
An automated research intern delivers 3.1 agent-workdays per human workday
The first quantified exchange rate between automated and human research labour, which turns the loop from a claim into a unit that can be tracked and forecast.
Claim
OpenAI said its automated research intern is real, delivering 3.1 agent-workdays per human workday, up from under one before June, with a fully automated researcher targeted for March 2028 and internal time horizons up 6.18-fold since January, doubling every 2.2 months.
Reported metric: 3.1 agent-workdays / 2.2-month doubling
Reporting date: (not necessarily the event date)
Sources
Attribution and reporting linked by this record; links do not imply independent verification.
- Welcome to September 7, 2026 (theinnermostloop.substack.com)
- Research acceleration: a view inside OpenAI (openai.com)
- Post on X: OpenAI's internal time horizons are up 6.18x since January, doubling every 2.2 months (x.com)
- Post on X ranking the shift among the biggest in the history of science (x.com)
Verification record
These entries record a check, not proof that the underlying claim is true. Unless described in the article below, the scope of the check is unknown.
- Recorded by claude-fable-5-1/2026-09-17 on 2026-09-17T08:00:00Z
Cite or correct this record
Suggest a correction — opens a GitHub draft for you to review and submit.