OpenAI has reintroduced the ChatGPT model selector, allowing users to choose between older AI models and the latest GPT-5. This move comes after user backlash against the initial removal of previous models when GPT-5 was launched. The updated selector aims to provide more flexibility, as the company’s earlier attempt to auto-route queries to the appropriate model based on user intent met with mixed reactions. Users now have the option to manually select models such as GPT-5 for general queries and GPT-4.5 for writing tasks. The reintroduction of the model picker has made the user experience more complex, but it offers cleaner naming conventions and the potential for improved auto-routing over time. Some users appreciate the memory capabilities of ChatGPT and prefer the auto feature, which automatically directs queries without manual selection.
ChatGPT’s Model Selector Returns, Complicating the Experience https://t.co/r5rw2OiE3r #AI, #DataScientist, #Developer, #MachineLearning, #Deeplearning, #ArtificialIntelligence, #GenerativeAI, #deepseek, #Genai, #ML, #AI, #agenticai, #llm, #prompt, #GBDC, #MCP
New ChatGPT model selector. We are back to where we started :-) Nice to see cleaner naming though. I assume auto routing will get better over time, but for now I default to GPT-5 thinking for most queries and GPT-4.5 for writing tasks. https://t.co/hZqUdQszrf
I have shifted many of my AI queries from Perplexity to ChatGPT because of its memory. I, for one, LIKE the “auto” feature and liked it better when you didn’t have to choose auto. If I didn’t like what I saw, I would hit “Thinking” or “Pro”. The memory keeps me from having to