PoTRE: Test-Time Reasoning inspired by Cognitive Heterogeneity

Researchers introduced PoTRE (Poly-Topological Reasoning Ensembles), a heterogeneous framework that decouples inference into four agents to improve complex reasoning in large language models. PoTRE achieves state-of-the-art accuracy of 49.92% on the Humanity's Last Exam (HLE) benchmark, surpassing the previous best official score. The framework uses a Task-Adaptive Aggregation Layer to dynamically reconcile perspectives from adversarial refinement, hierarchical planning, spectrum search, and direct chain agents.

arXiv:2607.20268v1 Announce Type: cross Abstract: While Large Language Models LLMs excel at many tasks, they frequently struggle with complex reasoning that requires long-horizon planning and iterative error correction. Furthermore, standard single-stream prompting proves brittle when models encounter novel abstractions or rigorous domain constraints. We introduce PoTRE Poly-Topological Reasoning Ensembles , a heterogeneous framework that decouples inference into four agents: 1 Adversarial Refinement Agent, 2 Hierarchical strategic Planning Agent, 3 Spectrum Search Agent, and 4 Direct Chain Agent. A final Task-Adaptive Aggregation Layer dynamically reconciles these perspectives -- via final candidate selection, semantic synthesis, or neuro-symbolic verification -- to produce a robust global solution. We evaluate PoTRE on three frontier benchmarks: ARC-AGI-2, Humanity's Last Exam HLE , and PRBench Finance. PoTRE achieves state-of-the-art accuracy of 49.92% on HLE, surpassing the previous best official score. We demonstrate that this architectural heterogeneity achieves improved reasoning performance using similar or fewer inference tokens compared to heavily scaled homogeneous baselines.