Definition
Long-horizon agents are agents that take useful action over many steps and long wall-clock times — running for hours or days on a single task. The frontier difficulties are credit assignment, error recovery, and keeping context coherent without overwhelming the model.
Episodes covering this
Worth reading next
Papers we haven't done a deep dive on yet, but would recommend on this topic.
- HIRL: A Human-in-the-Loop Benchmark for Agents that Know When to Ask for Help
- ELLM: Exploring with Large Language Models
- Cognitive Architectures for Language Agents
- OpenDeepSearch: Democratizing Search with Open-source Reasoning Models
- MemGPT: Towards LLMs as Operating Systems
- A-MEM: Agentic Memory for LLM Agents
- Generative Agents: Interactive Simulacra of Human Behavior
- Lost in the Middle: How Language Models Use Long Contexts