GPT — Field Evidence
paperUnverified
GPT · GPT · praise
Field note
OmniaBench: Benchmarking General AI Agents Across Diverse Scenarios — Large language models are increasingly evolving from text generators into general agents capable of understanding user requests, invoking external tools, and completing complex tasks through interaction. Howeve
collected 2026-07-17original 2026-07-16
Does this shift the US–China race?
Be the first to call it
Impact on the index
Benchmark win · minor
US +30China -7
Directional contribution — before recency decay and per-type diminishing returns. How it works →
Related Front
US vs ChinaLikely
U.S. FrontiervsChina Open-Weight
Likely — capability gap narrowing on common tasks
Letters from the Front
Be the first to file a report.