aiIMAGE
Paper: first retrieval step drives deep research agent performance
With the same retriever and agent loop, improving the first retrieval step raised GPT-5.5 from 83.1% to 90.5% on BrowseComp-Plus, with the gain coming mainly from the opening context.
elvis
@omarsar0Archived 2026.09.24Originally posted 2026.09.22

elvis @omarsar0Originally posted 2026.09.22
Impressive paper showing how much the first retrieval step matters for deep research agents. It helps to improve GPT-5.5 from 83.1% to 90.5% on BrowseComp-Plus with the same retriever and the same agent loop. It seems that the gain comes from the opening context. The authors https://t.co/niznBw9mJB
x.com/omarsar0/status/2102171057612566585Archived 2026.09.24 01:48