Story

On air. The free wire runs 15 minutes behind live. See pricing

World, crypto, finance and tech news, scored for impact.

CryptoTokenPostKorean, translatedFiled 14 h ago

46% success rate for tasks of 2-4 hours without intervention… 35% for 4-8 hours

Original headline2~4시간 과업 무개입 성공률 46%…4~8시간은 35%

Open the sourceReported by TokenPost, tokenpost.kr, 19 November 2026, by 김하린

Not investment advice. Muhabbit scores how much a story may matter, not which way to trade.

OpenAI’s coding agent completed 46% of research tasks estimated at two to four hours of human work without intervention in October, up from 23% in March, while no-intervention success on four-to-eight-hour tasks rose to 35% from 13%, according to the company’s “Research acceleration: The jagged frontier” report published on the 18th. The internal measurements, aggregated from August 1 to October 18, showed task-level no-intervention success of 79% for monitoring training progress, 35% for launching training, 29% for building evaluation, reinforcement learning and agent environments, and 25% for writing inference and serving code. The figures matter because they quantify how far OpenAI’s coding agent can operate autonomously on real research work rather than only short, well-scoped tasks, with performance varying sharply by task type. What to watch next is whether the upward trend in longer-horizon tasks continues and how the task-specific gaps, particularly in inference and serving code, narrow in subsequent reporting periods.

Muhabbit's AI wrote this summary in its own words from TokenPost's report and translated it from Korean. It can get details wrong, so check the source.

Named in this story

OpenAI