Online labor markets have experienced rapid growth, providing workers with a vast array of job opportunities. However, the unique challenges associated with task recommendation in these markets have not been fully addressed. Workers’ preferences are highly dynamic, influenced by factors such as skill development and changing interests over time. Moreover, the task listings are often short-lived and require quick matching with suitable workers. Existing recommendation approaches struggle to capture the complex and time-sensitive nature of worker-task interactions, leading to suboptimal task recommendations. We propose the time-weighted diffusion model for task recommendation in online labor markets in this paper. Utilizing the advanced capabilities of diffusion models to encapsulate intricate and dynamic relationships between workers and tasks, our model incorporates a unique weighting scheme that assigns higher importance to recent interactions. This enables the model to accommodate the changing preferences and abilities of workers, providing personalized and up-to-date recommendations. Experiments on a real-world dataset demonstrate the superiority of our approach, significantly outperforming state-of-the-art baselines.