video1.0<iframe src="https://www.loom.com/embed/2fdefa447c684340b7fd91383e47f89f" frameborder="0" width="1552" height="1164" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>11641552Loomhttps://www.loom.com11641552https://cdn.loom.com/sessions/thumbnails/2fdefa447c684340b7fd91383e47f89f-7a79246b0048d8e2.gif1169.294In this video, I discuss the question of what the best model is for AI applications, emphasizing that the answer is highly dependent on specific use cases and performance metrics. I showcase our AI Configs platform, which allows for dynamic model selection and experimentation, using a financial services chatbot as an example. Through A/B testing with models like Haiku 3.5, Nova Pro, and OpenAI's GPT-4-1, I highlight the importance of accuracy, response time, and cost, revealing that Nova Pro outperforms in secondary metrics while being significantly cheaper. I encourage viewers to consider a data-driven approach when selecting models, rather than relying solely on familiarity or perceived quality. Ultimately, I invite you to explore how LaunchDarkly can help you determine the best model for your specific needs.