← Back to resources

Benchmark

FutureSearch Deep Research Bench

Profile in progress

FutureSearch

Evaluate web-research agents on offline web-search tasks.

Open official page ↗

Resource profile

Access
Access status unknown
Access note
Separate project identity from USTC DeepResearch Bench.
Inputs
Agent search/reasoning outputs over each offline web snapshot.
Outputs
Task scores against curated answers.
License
License not yet verified
Limits
Retro results use frozen web snapshots; live commercial-tool results are separately identified.
Evaluation conditions
91 tasks, each with 10k–100k stored webpages; do not combine its scores with USTC DeepResearch Bench.
Code release
Not verified
Data release
Not verified
Site execution check
Not verified

Associated timeline

No verified associated update is currently displayed.