Agents
OpenResearcher: a reproducible and scalable pipeline for training deep research agents
A deep research agent has to do more than answer a question. It plans, searches, opens documents, gathers evidence, reasons across sources, and keeps going across dozens or hundreds of tool calls befo
A deep research agent has to do more than answer a question. It plans, searches, opens documents, gathers evidence, reasons across sources, and keeps going across dozens or hundreds of tool calls before it lands on an answer. Training one means teaching all of that. That's where most teams hit a wall.
Related
- Open-source agent for long-horizon deep research https://github.com/TIGER-AI-Lab/OpenResearcher
- AgentFlow: when the agent's workflow learns
- Multi-Turn Evaluation of Deep Research Agents Under Process-Level Feedback
Source: Lambda Labs | 2026-09-10