I can’t skip reasoning model & reasoning model used even it’s not necessary (very frustrating)

Enhancing User Experience in AI-Assisted Chat Platforms: Addressing Model Selection and Reliability Challenges

In the evolving landscape of AI-driven conversational tools such as ChatGPT, users often encounter certain frustrations related to model selection and the system’s ability to accurately handle external data sources. These issues, while seemingly technical, significantly impact the overall user experience and trustworthiness of the platform.

Difficulty Skipping Unnecessary Reasoning Models

One common concern is the inability to bypass specific processing models, such as the GPT-5 Thinking Mini model, which the system automatically selects as a default. Users frequently express a preference to continue using this model but find themselves unable to do so effortlessly. During the initial response, the absence of a “Skip” button necessitates waiting approximately 30 seconds, leading to impatience. Furthermore, when opting to retry a response—by clicking the “Retry” option at the bottom of a message—the model is sometimes still pre-selected, creating redundancy and frustration.

Recommended Improvements:
– Ensure that users can always skip or select their preferred reasoning model at any point, including the initial response.
– Enhance the system to allow persistent user preferences to use a specific model (for example, consistently using GPT-5 Mini).
– Optimize response flow to minimize waiting times and streamline user interactions.

Addressing Model Performance and User Preferences

Beyond skip functionality, there’s a need to improve the reasoning models themselves. Users have observed that these models occasionally drift off-topic or take responses that are overly formal or serious, diminishing engagement and entertainment value. Providing options for users to set or lock their preferred models can enhance satisfaction and customization.

Ensuring Accurate External Data Integration

Another critical aspect pertains to the system’s handling of external data sources, such as website browsing. Users have noted instances where ChatGPT claims to have visited or summarized websites but subsequently provides fabricated or hallucinated content. For example, when asked to summarize a website about “how cookies are good,” the AI may instead generate unrelated or inaccurate information, such as a story about pumpkin pie. While occasional browsing failures are understandable, the lack of transparency about these failures hampers trust.

Actionable Suggestions:
– Clearly inform users when website browsing or external data retrieval fails, rather than presenting potentially inaccurate summaries.
– Improve the AI’s ability to recognize and communicate limitations in accessing external content, maintaining transparency and honesty.

Conclusion: Striving for Trustworthy and User-Centric AI Tools

Ultimately, the goal should be to create AI chat systems that are reliable, transparent,