While the global climate crisis creates ecological, economic and social pressures on urban built environments, the discipline of landscape architecture develops adaptation strategies centering on nature-based solutions to combat these pressures. Reconstructing urban landscape areas in accordance with environmental resilience and sustainability principles requires expressing complex datasets in an aesthetic and understandable language. The main objective of this research is to comparatively examine the performances of generative artificial intelligence models capable of text-to-image conversion in the visual simulation of climate crisis adaptation strategies in the urban landscape, in light of computational metrics and statistical analyses. In this study, which adopts an experimental and quantitative research design, a total of 500 images were generated from both algorithms for five different designated climate adaptation strategy scenarios. The obtained images were analyzed using Structural Similarity Index (SSIM), Learned Perceptual Image Patch Similarity (LPIPS) and Contrastive Language-Image Pre-training (CLIP) Score algorithms. In the statistical analysis of the data, the non-parametric Mann-Whitney U test and the Independent Samples T-Test to verify the sub-distributions of the dataset were applied and effect sizes were formulated. The findings revealed that the Midjourney v6 model was statistically significantly superior to DALL-E 3 in LPIPS and SSIM scores, while the DALL-E 3 model showed significant success in responding to complex textual prompts as measured by the CLIP score. In conclusion, it has emerged that generative artificial intelligence tools can be integrated into urban landscape design processes as complementary decision support systems depending on the context of the subject and the stages of the design process.