Depending on what and how caching helpers cache content in their finite storage, the caching helpers can offer either a content diversity gain by serving diverse content or a cooperative gain by jointly transmitting the same content. This paper identifies a tradeoff between the content diversity gain and the cooperative gain according to content placements and proposes a probabilistic content placement to optimally balance the tradeoff. Using stochastic geometry, we quantify this tradeoff by deriving the cache hit rate and the rate coverage probability. To efficiently control the tradeoff, we determine the near-optimal caching probabilities that maximize the average content delivery success probability with the cooperative caching helpers. Our analysis and numerical results reveal that our proposed content placement outperforms the conventional caching schemes, such as caching with uniform probabilities, caching the most popular contents, and caching the content maximizing the cache hit, in terms of the average content delivery success probability.