NVIDIA Nemotron 3 Embed 8B tops Perplexity's new Q2D-Web agentic RAG benchmark

JFPuget · x · 2026-09-11

Perplexity introduced Q2D-Web (Query2Doc-Web), a benchmark and public leaderboard for evaluating retrieval in agentic RAG systems, built on 190M web documents and nearly 70K agent-reformulated queries across 10 languages. NVIDIA announced its Nemotron 3 Embed 8B ranks #1 in combined nDCG@10 on the benchmark, the first large-scale web retrieval eval focused on agent-reformulated queries.

Original post →

More from Models

Models channel →