OpenAI Brings Back Model Selection After GPT-4o Backlash
After criticism, Openai brings the model selection for Chatgpt. GPT-4O is available again after the surprising shutdown caused resentment. There are also new GPT-5 options that should offer more flexibility.
Chatgpt model selection returns
Openai has reintroduced the model selection for Chatgpt after GPT-5 was introduced as a new standard model for all users. After massive criticism, the company again made GPT-4O available for all paying users. CEO Sam Altman promised that Openai would inform you in good time when the model will be shut down.
The decision marks a U-turn in the strategy of Openaai: originally the company had sought a uniform AI experience in which a single, advanced model was to cover all user needs. However, this vision proved to be too ambitious in view of the diverse applications and preferences of users. How Techcrunch reports, the mood in the Reddit community after the GPT-5 launch was clearly negative. In particular, users who use chatt for creative cooperation, emotional nuances or role -playing games felt impaired by the changes.
New options for GPT-5
Users of paid tariffs can now choose between different GPT 5 variants: car, fast and thinking. The car setting works like an intelligent router based on the type of conversation, complexity and explicit instructions automatically selects the right model.
This technology is continuously trained with real signals, including user change between models, preference rates for answers and measured correctness. GPT-5-Thinking is available plus and team users with a limit of 3000 messages per week, after which it is automatically switched to GPT-5-Thinking Mini. This variant is particularly designed for complex problem solutions and analytical tasks, but requires significantly more computing resources than the standard models.
Access to Legacy models
Additional older models such as O3, O4-Mini and GPT-4.1 can be activated via the settings. GPT-4.5 is only reserved for pro users because it is very resource-intensive. These changes only affect Chatgpt consumers, while the API continues to offer access to the old models. The availability of different model generations reflects the different development phases of Openais Ki technology.
Each model has been optimized for specific applications and brings its own strengths and weaknesses. GPT-4O, for example, is considered particularly creative and empathetic, while GPT-5 is designed for accuracy and fidelity.
