Intel / post

We're introducing Q2D-Web (Query2Doc-Web), a benchmark and public leaderboard…

Source: x.com
Date: Published Sep 9, 2026

We're introducing Q2D-Web (Query2Doc-Web), a benchmark and public leaderboard for evaluating retrieval in agentic RAG systems. Q2D-Web tests how embedding models perform on large-scale web search using agent-reformulated search queries. Read more: https://t.co/s476SxkE1L

Sep 9, 2026 View on X

Context

Perplexity introduced Q2D-Web (Query2Doc-Web), a new benchmark and public leaderboard for scoring how well models retrieve relevant documents inside agentic systems, retrieval-augmented generation setups where an searches for source documents before answering. The distinguishing feature is the query set: instead of using queries typed directly by people, Q2D-Web tests retrieval using large-scale web search queries that were reformulated by an AI agent.

That distinction is meant to reflect how agents actually search in practice, since an agent often rewrites or expands a query before running it, in ways that differ from how a person would type the same request. NVIDIA was among the first to report a result on the new benchmark, ranking its Nemotron 3 Embed model first for combined nDCG@10, giving the leaderboard an early public data point shortly after launch.

Terms in this piece · Glossary

embedding — A list of numbers representing a piece of text's meaning, so that similar meanings end up numerically close and can be searched.
RAG — Retrieval-augmented generation — fetching relevant documents first and pasting them into the model's context so it answers from your data instead of memory.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.

Read the source x.com