ChatGPT now decides — GPT-5 automates model selection

OpenAI’s GPT-5 uses a smart router to choose models by task complexity, delivering faster and more reliable responses. Learn how GPT-5’s unified system routes prompts to different models to optimize performance and response quality.

TidBits explained new router model in GPT-5.

GPT-5 simplifies ChatGPT

ChatGPT has long defaulted to GPT-4o, which is fast and inexpensive but provides weaker responses than more expensive reasoning models, such as OpenAI o3. These models can take 10–15 minutes to generate a response. GPT-5 allows ChatGPT to analyze your prompt and decide which of several models of various sizes and abilities to use.

If you have never used any other model than GPT-4o, GPT-5 could lead to much better responses. On the other hand, if you carefully select your models, you might be frustrated if GPT-5 chooses differently in a specific case. To accommodate these users, OpenAI allows you to manually select GPT-5 or the slower, more capable GPT-5 Thinking from the model dropdown menu at the top. Those pining for GPT-4o can access it by turning on Show Legacy Models in ChatGPT’s settings.

"A surprising number of people have never seen what AI can actually do because they’re stuck on GPT-4o, and don’t know which of the confusingly-named models are better," professor Ethan Mollick said.

GPT-5 eliminates this issue by automatically selecting models for you. Rather than being one model, GPT-5 is a switch that selects from multiple GPT-5 models of various sizes and capabilities.

Read more:

ChatGPT to stop acting like a therapist in OpenAI’s new update

Top complaints about ChatGPT-5 — what users hate

OS 26 will use ChatGPT-5 to boost Siri and Apple Intelligence