# GATS Cuts LLM Calls in Agent Planning With Graph-Augmented Tree Search

The method pairs layered world models with graph-augmented search to reduce the heavy per-step language-model inference that slows LATS and ReAct.

- Published: 2026-07-13T05:34:23.811Z
- Canonical: https://polylog.news/ai/2026-07-13/gats-cuts-llm-calls-in-agent-planning-with-graph-augmented-t
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [arXiv](https://arxiv.org/abs/2607.08894)

A new preprint, GATS: Graph-Augmented Tree Search with Layered World Models, addresses a practical bottleneck in language-model agents. Existing planners such as Language Agent Tree Search (LATS) and ReAct rely heavily on model inference du…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-07-13/gats-cuts-llm-calls-in-agent-planning-with-graph-augmented-t (subscription information: https://polylog.news/pricing).