With the rapid development of Internet technology, people's lives are increasingly embedded in online communities. The implementation scenarios of green behavior are extended from offline to online, and individual decision-making is more likely to be influenced by others in the group. However, little is known about the decision-making process of group participation in online green behavior (OLGB). To fill this gap, we took the online environmental project Ant Forest as an example and constructed a multi-agent simulation model incorporating members' gratification types, social network connectivity, and external incentives. Then, we analyzed the decision-making characteristics of groups participating in OLGB. The results showed that: (1) When the ratio of members seeking entertainment, social, and achievement gratifications was 3:3:4, more people chose to accept OLGB and it has the most stable decision status; (2) Under low network connectivity, the group was more inclined to accept OLGB, and the decision status was more stable; (3) When high-intensity financial incentives and medium-intensity non-financial incentives were coupled, the number of people choosing OLGB was the largest, and the decision-making status was the most stable. These findings provide suggestions for policymakers and online green project operators on how to promote more people to engage in OLGB.