The HIPE-2026 campaign addresses the challenge of extracting person-place relations from noisy, multilingual historical documents. Moving beyond previous editions focused on named entity recognition, this third iteration targets temporally grounded relationships labeled as 'at' and 'isAt'. The evaluation involved 17 participating teams processing data in French, German, and English across three distinct datasets. These datasets comprised nineteenth and twentieth-century newspaper text alongside a surprise domain set of early modern French literary works. A key feature of the campaign was its three-fold framework assessing predictive accuracy, computational efficiency, and cross-domain generalization. Results from over 40 submitted runs demonstrated a wide variety of strategies, ranging from large language models to lightweight classifiers. The findings highlight the inherent trade-offs between accuracy, efficiency, and robustness in large-scale historical relation extraction.