Diffusion-KTO is a novel framework for aligning text-to-image diffusion models with human preferences using only per-sample binary feedback. Unlike previous methods that require pairwise preference data or complex reward models, Diffusion-KTO directly optimizes the model using per-image binary feedback signals, such as likes or dislikes. This approach allows for more efficient and scalable alignment with human preferences. After fine-tuning using Diffusion-KTO, text-to-image diffusion models show improved performance compared to existing techniques, including supervised fine-tuning and Diffusion-DPO, in terms of both human judgment and automatic evaluation metrics like PickScore and ImageReward. Diffusion-KTO extends the utility maximization framework from KTO to diffusion models, enabling the model to learn from per-image binary feedback and align with human preferences more effectively. The framework avoids the need for expensive pairwise preference data and allows for broader applicability in aligning text-to-image diffusion models with human preferences. Diffusion-KTO demonstrates significant improvements in image preference and image-text alignment when evaluated by human judges and automated metrics. The method is implemented using the Kahneman-Tversky utility function, which is effective in aligning models with human preferences. Diffusion-KTO outperforms existing methods in terms of alignment with human preferences and shows strong results in both qualitative and quantitative evaluations. The framework is applicable to a wide range of text-to-image generation tasks and can be used to cater to specific user preferences. The code for Diffusion-KTO is available at https://github.com/jacklishufan/diffusion-kto.Diffusion-KTO is a novel framework for aligning text-to-image diffusion models with human preferences using only per-sample binary feedback. Unlike previous methods that require pairwise preference data or complex reward models, Diffusion-KTO directly optimizes the model using per-image binary feedback signals, such as likes or dislikes. This approach allows for more efficient and scalable alignment with human preferences. After fine-tuning using Diffusion-KTO, text-to-image diffusion models show improved performance compared to existing techniques, including supervised fine-tuning and Diffusion-DPO, in terms of both human judgment and automatic evaluation metrics like PickScore and ImageReward. Diffusion-KTO extends the utility maximization framework from KTO to diffusion models, enabling the model to learn from per-image binary feedback and align with human preferences more effectively. The framework avoids the need for expensive pairwise preference data and allows for broader applicability in aligning text-to-image diffusion models with human preferences. Diffusion-KTO demonstrates significant improvements in image preference and image-text alignment when evaluated by human judges and automated metrics. The method is implemented using the Kahneman-Tversky utility function, which is effective in aligning models with human preferences. Diffusion-KTO outperforms existing methods in terms of alignment with human preferences and shows strong results in both qualitative and quantitative evaluations. The framework is applicable to a wide range of text-to-image generation tasks and can be used to cater to specific user preferences. The code for Diffusion-KTO is available at https://github.com/jacklishufan/diffusion-kto.