Streaming Communication in Multi-Agent Reasoning

Benchmark Model Rank Results
mathematical-reasoning-on-aime-2025STREAM-MA (Claude Opus 4.6, Graph)#8Accuracy: 95.83
question-answering-on-humanity-s-last-exam-hleSTREAM-MA (Claude Opus 4.6, Graph)#17Accuracy: 27.68
mathematical-reasoning-on-aime-2026STREAM-MA (Claude Opus 4.6, Chain)#20Accuracy: 89.58
multiple-choice-question-answering-mcqa-on-gpqa-diamondSTREAM-MA (Claude Opus 4.6, Chain)#24Accuracy: 87.37