Generating high-quality textures for 3D meshes from limited input remains a challenging problem in computer graphics. Existing text-driven methods often struggle with multi-view consistency and geometric fidelity, especially when relying on text prompts. We propose V2Tex, a novel framework that leverages video diffusion models to synthesize detailed and seamless textures from text. V2Tex first generates the primary view using a diffusion model, which serves as the input for a video diffusion model. By leveraging the 3D spatial awareness of video generation models, we further finetune them to generate surround-view video sequences. Subsequently, we design a depth prior-guided surround video generation method to enhance geometric consistency and structural accuracy in the generated videos. The generated frames are then projected onto the mesh's UV space to create a complete, high-quality texture map. Our method bypasses the need for complex optimization procedures or explicit seam removal, offering a simplified and efficient solution. Extensive experiments demonstrate that V2Tex significantly accelerates the texture generation process compared to existing state-of-the-art methods, while producing textures of comparable or superior quality. Moreover, our approach simplifies the texturing process, making it applicable to scenarios where acquiring multiple views or complex 3D data is impractical. V2Tex opens exciting new possibilities for text-driven 3D texturing, paving the way for more intuitive and efficient content creation pipelines.