In this research, we introduce a Generative AI-driven storytelling system that uses user prompts to produce a story with logic and without any grammar mistakes. The system uses DistilGPT2 and GPT2 transformer-based models and selects the best output using the BERTScore and coherence score. The chosen story is further refined using a grammar correction model based on T5. The final production is well-structured with clear paragraphs and natural dialogue. This approach improves the quality of the story for educational, entertainment, and creative writing. The evolution of Generative AI has opened new avenues in creative writing and content generation. This "GenAI-driven Storyteller" project presents an intelligent system that automatically generates engaging and grammatically correct short stories based on user-provided prompts. The system integrates the strengths of multiple language models—GPT2 and DistilGPT2—to produce diverse narrative outputs. These stories are evaluated using BERTScore for semantic relevance and a coherence scoring mechanism to measure logical flow. A weighted ensemble approach determines the most compelling version among the generated outputs. The selected story is then refined for grammar and fluency using a fine-tuned Google T5-small model. This hybrid architecture ensures that the final story is not only contextually aligned with the user's prompt but also polished in structure and readability. The project highlights the potential of combining generation, evaluation, and correction models to enhance automated storytelling systems.