Why does the AI model name shown in my custom template look different from the one I selected?
In some older custom-template cases, if the template asked the summary to directly display the “AI model used,” the model name shown in the output could be inaccurate or inconsistent.
Does this mean my selected model was not actually used?
Not necessarily. In most cases, this issue only affected the model name shown in the output text, not the model that was actually selected for the task.
Why did this happen?
When a template asks the model to state its own name in the generated content, the answer may be influenced by context and generation behavior, which can cause the displayed model name to be unstable or incorrect.
Has this issue been fixed?
Yes. This scenario has been improved, and when a custom template asks to display the model used for a run, the shown result should now match the model actually selected.
Does this affect all summary results?
No. The confirmed issue was limited to cases where a custom template explicitly asked the output to include the model name.
Can I use the model name shown in the output as proof of which model was used?
For general reference, yes, but for verification, troubleshooting, or auditing, the most reliable source is the server-side invocation record rather than the generated text alone.
What should I do if I still see this issue?
Please contact support and provide the following information:
- The template content
- The model you selected
- A screenshot of the result
- The platform you used (App or Web)
- When the issue happened
- Whether the issue can be reproduced consistently
- Any related public post or link, if available