The uncertainty of cloud resources makes task scheduling very difficult, and may results in a waste of energy of cloud resources if some cloud resources stay in unusable state for a long time. This paper proposes the cloud resource evaluation model based on entropy optimization (EOEM) to solve the uncertain problem. The evaluation results also can be used to develop resource management strategies, which will be sufficient for saving energy. Particularly, the proposed model may release long-term unavailable resources to reduce energy consumption. With the entropy increasing minimum principle, the proposed model takes the objective function of resource providers and user as constraints of maximum entropy function, thus it can balance profits of both cloud resource providers and cloud users. Meanwhile, the system utilization can be maximized and the resources satisfying QoS can be screened. For demonstrating their performance we evaluate the proposed model by simulation experiment. Simulation results reveal the positive effect of the entropy optimization to satisfaction degree of users and completion time of task. The model also shows better adaptability when cloud resources join or quit the cloud.