--- schema: formation.pattern/v0.1 kind: pattern visibility: public aliases: - 'Language Agent Tree Search' - LATS - 'Monte Carlo tree search agent' - 'best-first tree search' - 'Tree of Thoughts' - 'beam search over reasoning steps' canonical_url: https://topologyindex.com/patterns/tree_search.md description: 'Partial attempts branch into a tree, an evaluator scores the branches, and the most promising one is extended next.' executable_here: false executor_requirements: - routing_on_unmeasured_conditions - aggregation_over_agent_opinions family: adaptive findings_page: /patterns/tree_search/findings.md observed_in: [] path: /patterns/tree_search.md pattern: tree_search pattern_index: /patterns/index.md product_api_version: v1 published_findings: - benchmarks: - HumanEval - MBPP - HotpotQA - WebShop - 'Game of 24' compared_against: 'ReAct, Reflexion, and the tree-search methods Tree of Thoughts and RAP' direction: helped source_id: arxiv:2310.04406 task_domain: 'Programming, interactive question answering, web shopping and math puzzles' task_domains: - coding - reasoning - operations url: https://arxiv.org/abs/2310.04406 - benchmarks: - 'SWE-bench Lite' compared_against: 'The same open-source software agent without tree search' direction: helped source_id: arxiv:2410.20285 task_domain: 'Repository-level software engineering: resolving GitHub issues' task_domains: - coding url: https://arxiv.org/abs/2410.20285 - benchmarks: - VisualWebArena - WebArena compared_against: 'The same web agent without search' direction: helped source_id: arxiv:2407.01476 task_domain: 'Web automation on realistic websites' task_domains: - operations url: https://arxiv.org/abs/2407.01476 - benchmarks: - 'Game of 24' - 'Creative Writing' - 'Mini Crosswords' compared_against: 'Input-output and chain-of-thought prompting of the same model' direction: helped source_id: arxiv:2305.10601 task_domain: 'Puzzles and constrained creative writing that need planning or search' task_domains: - reasoning url: https://arxiv.org/abs/2305.10601 - benchmarks: - MATH compared_against: 'Best-of-N sampling scored by the same verifier' direction: mixed source_id: arxiv:2408.03314 task_domain: 'Competition mathematics' task_domains: - reasoning url: https://arxiv.org/abs/2408.03314 qualifying_evidence: [] references: - https://arxiv.org/abs/2305.10601 - https://arxiv.org/abs/2310.04406 - https://arxiv.org/abs/2407.01476 schema_version: v0.1 title: 'When to use the tree_search multi-agent pattern (Language Agent Tree Search, LATS)?' unmet_executor_requirements: - routing_on_unmeasured_conditions - aggregation_over_agent_opinions --- # When to use the tree_search multi-agent pattern (Language Agent Tree Search, LATS)? **Short answer:** test it only when early decisions matter, a wrong turn can be recognised before the end, and the environment can be put back into an earlier state; otherwise start from the [decision guide](/patterns/index.md) row for your task. A hypothesis, not a ranking; [published findings](/patterns/tree_search/findings.md) keep the unfavourable ones. The `tree_search` multi-agent pattern (family `adaptive`): partial attempts branch into a tree, an evaluator scores the branches, and the most promising one is extended next. Also called: Language Agent Tree Search, LATS, Monte Carlo tree search agent, best-first tree search, Tree of Thoughts, beam search over reasoning steps. ## The arrangement - Structure: An acting role that proposes next steps, an evaluating role that scores the states they reach, and a search rule over the tree of states. - Control: The search rule picks which state to extend from the scores, and can go back to an earlier state when a branch looks worse than another. - Shared state: The tree: every state explored, its score and the path back to the start. ## When it is hypothesised to fit Hypothesised to fit when early decisions matter, a wrong turn can be recognised before the end, and the environment can be put back into an earlier state. A hypothesis to test on a workload, not a finding: this service has not tested it. ## Failure modes to watch for - The evaluator steers the whole search; a scorer with a blind spot sends every branch the same wrong way. - Going back needs a way to restore an earlier state, which a live system, a shared database or an irreversible action does not offer. - Cost grows with the branching and the depth, and a search that fails has still paid for every state it expanded. - On easy tasks a strong search can do worse than plain sampling, because it optimises the evaluator’s mistakes. These are things to watch for, not outcomes anyone measured here. ## What published studies found Attributed to each source and stated without figures; unfavourable results are included on purpose, and none of this is evidence produced by this service. What each was compared with is in `published_findings`; each finding in words, with its caveat, is at [/patterns/tree_search/findings.md](/patterns/tree_search/findings.md). Reviewed 2026-09-23. Each label says how this pattern fared against what it was compared with, as the source reports it; the labels are defined at [/docs/schemas/pattern/v0.1.md](/docs/schemas/pattern/v0.1.md). - **Helped**: [Language Agent Tree Search Unifies Reasoning Acting and Planning in Language Models](https://arxiv.org/abs/2310.04406) - **Helped**: [SWE-Search: Enhancing Software Agents with Monte Carlo Tree Search and Iterative Refinement](https://arxiv.org/abs/2410.20285) - **Helped**: [Tree Search for Language Model Agents](https://arxiv.org/abs/2407.01476) - **Helped**: [Tree of Thoughts: Deliberate Problem Solving with Large Language Models](https://arxiv.org/abs/2305.10601) - **Mixed**: [Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters](https://arxiv.org/abs/2408.03314) ## Sources that describe it Where the arrangement is described. A source is a reason to test it, never a result for it. - [Tree of Thoughts: Deliberate Problem Solving with Large Language Models (Yao et al.)](https://arxiv.org/abs/2305.10601) - [Language Agent Tree Search Unifies Reasoning Acting and Planning in Language Models (Zhou et al.)](https://arxiv.org/abs/2310.04406) - [Tree Search for Language Model Agents (Koh et al.)](https://arxiv.org/abs/2407.01476) ## Can this deployment execute it No. The runner does not provide everything this arrangement needs. What is missing: - `routing_on_unmeasured_conditions` (the next role chosen from conditions the executor does not measure): A condition outside the measured signal set is refused as an unsupported capability, and every adaptation rule in an execution lock is refused with it. - `aggregation_over_agent_opinions` (a rule that combines what several roles judged into one outcome): The supervisor decides only from measured signals, so it cannot read a verdict, weigh it or count votes. The runner gives each episode one fresh workspace and only moves forward through declared roles: it cannot save a state, return to it and try another branch, and it does not read a scorer’s judgement. A search scored by the public checks would still need those saved states. A topology formation that declares what this deployment cannot run is still stored, and is refused with an unsupported-capability error rather than reduced to a simpler arrangement and run anyway. ## Evidence for this pattern None. No qualifying evidence exists here for this pattern, and by design none ever will: an evidence snapshot is about the complete execution configuration that produced the observations, so the most it can support is a claim about that configuration on that workload. ## Observed in No research record on this service reports observing this arrangement. ## How evidence works here A pattern names how work is organized, never how well that works. Evidence in this service attaches to a complete execution configuration — a topology formation together with an execution lock pinning models, tools and runtime — and never to a pattern on its own, so no page here carries a success rate, a cost or a run count for an arrangement.