GPT-5 Is Here: The One Feature That Changes Everything

OpenAI’s GPT-5 is now live, and while I didn’t receive an embargo—partly due to my occasional critical stance toward the company—I’ve reviewed every major early assessment: from Ethan Mollick, Tyler Cowen, Every, Latent Space, METR, Artificial Analysis, and others. I’ve watched the full one-hour demo, studied the four blog posts, and examined the model’s system card PDF. The core upgrades in GPT-5 can be summed up in a single paragraph: it’s faster and more affordable than its predecessors, with performance that now dominates the price-performance curve—making it the most efficient model available, even if some feel the leap isn’t as dramatic as hoped, especially when compared to Anthropic’s Claude 4 and 4.1. It’s accessible to nearly everyone, with OpenAI’s new system automatically selecting the best internal model for your task—without user control. It excels at tool use, enabling more sophisticated agentic behavior, including end-to-end research workflows. It’s the strongest coding model yet, a crucial advantage that helps OpenAI’s own engineers accelerate development of GPT-6. It still fails the ARC-AGI 2 benchmark, and it remains notably weak at creative writing—still struggling with originality, tone, and emotional depth. These points are consistent across all the sources I reviewed. But there’s one aspect that hasn’t been highlighted enough—and it’s a strategic win for OpenAI. GPT-5 doesn’t just improve performance; it reinforces OpenAI’s control over the user experience. By automatically routing queries to the most appropriate internal model, OpenAI is effectively abstracting the model selection process from users. This reduces confusion, streamlines delivery, and gives OpenAI greater leverage over how the model is used. It’s not just about speed or accuracy—it’s about centralizing intelligence, not just in capability, but in decision-making. This subtle shift matters. It means users don’t need to understand model differences. They just get the best result, as determined by OpenAI’s internal logic. That’s not just a technical upgrade—it’s a strategic move to deepen user dependence on the platform, tighten product control, and lay the groundwork for future AI systems that operate with even less human oversight. In short, GPT-5 isn’t just better at what it does—it’s better at how it does it, and more importantly, how it keeps users within its ecosystem.