Definition
Plain language
An open-source search-agent system trained mostly with supervised fine-tuning on long, diverse trajectories.
As stated in the literature
An open-source search agent (OpenSeeker-v2) that achieves frontier-comparable BrowseComp performance via SFT on hard, tool-diverse trajectories without an RL stage.
Also called: OpenSeeker-v2
Why it matters: It's evidence that careful SFT on diverse trajectories can rival RL-trained search agents, simplifying training pipelines for many teams.
For example, OpenSeeker-v2 learns its search behavior by imitating long, multi-tool browsing trajectories rather than via reinforcement learning.
Heard on the show
“… The paper is called "OpenSeeker-v2: Pushing the Limits of Search Agents with Informative and High-Difficulty Trajectories," and …”Episode 021 — Ten Thousand Examples Beat the Full Industrial Pipeline for Search Agents