CRITICAL BUG: 'Follow-up Questions' Feature Fails, Displays 'gpt-5' Error with Standard Model

[sharpdima]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
• [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:)
• Please do not modify this template :) and fill in all the required fields.

Content

I am experiencing a critical bug related to the "Follow-up Questions" feature, which persists across different environments.

Key Findings & Evidence:

Direct Causation: The error only appears when the "Follow-up Questions" feature is activated in the Features panel. Disabling this feature immediately removes the error.

Incorrect Model Call (The Bug): After the model response is displayed, the following error appears in the UI: Model openai/gpt-5 credentials is not initialized.

No GPT-5 Usage: I confirm that I am NOT using, nor have I ever configured, any GPT-5 model. My active model is a standard commercial model (e.g., gpt-3.5-turbo or a custom-compatible model like gpt-4.1-mini).

Issue Isolation: This feature also fails to display suggested questions on another VPS running a different Dify installation, even though no error message is shown there, confirming a fundamental issue with the feature's logic.

Conclusion: The Dify system's code for "Follow-up Questions" is incorrectly defaulting to, or hardcoding, an obsolete/unreleased model name (gpt-5) during its initialization check, leading to a parameter mismatch error and feature failure.

Request: Please investigate this core bug in the Follow-up Questions feature logic and provide the specific version updates or backend configuration fixes required to resolve the erroneous gpt-5 call.

Thank you for your urgent attention to this matter.