Query Agent Introduces Effort Tiers, Massively Boosting AI Search Performance
CShorten30 · x · 2026-08-13
The developer has integrated Scaling Test-Time Compute into the Query Agent's Search Mode, introducing three new effort tiers: Medium, High, and Ultrahigh.
Benchmarked across 8 reasoning-intensive and domain-specific retrieval datasets, all effort tiers outperformed traditional Hybrid Search. The Ultrahigh tier delivered exceptional results, roughly quadrupling Hybrid Search’s nDCG@10 on BRIGHT Biology and achieving a 7x improvement in Success@1 on OBLIQ-Bench Congress.
More from coding & agent
- AI Digital Coworkers Emerge as New Meta: OpenAI, Grok and Others Join In — tushaarmehtaa · 2026-08-13
- Grok 4.6 in Action: 10 Wild Coding and 3D Generation Demos Within 48 Hours — minchoi · 2026-08-13
- Microsoft Paper Reveals Hidden Costs of Bad Skills in AI Agent Harnesses — omarsar0 · 2026-08-13
- Stop Waiting for Better Models: Architecture is the Key to Agent Reliability — import_jmr · 2026-08-13
- NVIDIA Offers 7 AI Courses for Free, Covering Agentic AI and Robotics — ifioknkem · 2026-08-13
- Claude Code's harness makes almost everything worse, says user — aiamblichus · 2026-08-13