Frontier models reach only about 50-54% Pass@8 when an answer needs both SQL and web search.

Snowflake's HybridDeepResearch supplies 380 tool-dependent tasks grounded in LiveSQLBench-Base-Lite databases plus public web corpora, testing whether an agent preserves constraints while moving evidence between systems. GLM-5.2, Claude Sonnet 4.6 and GPT-5 all cluster around that number on the hard subset, and directional reasoning proves substantially harder than parallel intersection. Existing deep-research benchmarks test the open web or structured data in isolation and never measure the handoff, which is where production agents actually break.