Nace AI Releases Drex for Search for AI Agents: Hits 92.05% on BrowseComp-Plus
Nace AI launches Drex for Search, achieving 92.05% on BrowseComp-Plus with an average of 12.6 search calls for document retrieval.
Stock photo for illustration only, not from the actual event
- Nace AI releases Drex for Search for AI agents
- Achieves 92.05% benchmark score on BrowseComp-Plus
- Requires an average of only 12.6 search calls
- Reduces agent turns on OfficeQA Pro
Nace AI has officially introduced Drex for Search, bringing its advanced Drex decision models into document search capabilities specifically tailored for AI agents. The new release aims to enhance how autonomous agents retrieve and process complex information.
According to Nace, the system achieves a remarkable 92.05% score on BrowseComp-Plus while utilizing an average of just 12.6 search calls per task, demonstrating high accuracy and efficient information filtering.
Stock photo for illustration only, not from the actual event
In addition to strong benchmark scores, Drex for Search successfully reduces the number of agent turns required on OfficeQA Pro, allowing AI systems to complete queries with fewer steps and greater operational efficiency.
Developing specialized decision models for document search addresses a major bottleneck in autonomous AI agents, which often struggle with redundant search loops. Nace AI's approach of keeping search calls down to 12.6 while scoring above 92% highlights significant progress in algorithmic efficiency and token optimization for agentic workflows.
Source: MarkTechPost
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment