As climate change accelerates, it is critical to reduce CO2 emissions equitably and efficiently. However, a unified and comparable framework for quantifying and identifying city-to-city collaborative CO2 emission reduction (CLCER) relationships remains lacking. This gap hampers cross-city comparability, robust source–sink identification, and evidence-based allocation of collaborative reduction responsibilities, limiting the design of fair and efficient regional policies. This study examines the spatial–temporal patterns of CLCER among cities in the Yangtze River Economic Belt from 2010 to 2019. To capture both economic interactions and physical transmission mechanisms, a composite CLCER index was developed by integrating an improved gravity model with Fick’s Law of diffusion. This dual “gravity–diffusion” framework, which incorporates linear programming and the CLCER index, enables systematic identification of carbon source and sink cities while providing a balanced representation of spatial interaction and diffusion path mechanisms. The findings revealed that: (1) From 2010 to 2019, the total CLCER demand increased from 483.85 Mt to 503.52 Mt, reflecting the rising demand for cross-city collaboration driven by the strengthened differentiation of carbon source and sink cities; (2) Cities in the Central Region and Western Region, apart from cross-city offset within their own regions, accounted for 99.43 % of the Eastern Region’s CLCER demand during the study period; (3) City-level CLCER shifted towards long-distance collaboration, with average inter-city CLCER increasing from 3.75 Mt to 3.90 Mt, and the average collaborative distance expanding from 851.44 km to 983.82 km; (4) CLCER spatio-temporal patterns were characterized by growing source–sink differentiation, stronger cross-regional linkages, and more dispersed inter-city connections. This study provides insights into the relationships, degrees, and emerging patterns of CLCER, offering a framework for achieving fair and efficient regional emission reductions.