Description
Description:
When using a GPT-5-Mini model (via a custom provider) as the model for prompt enhancement or title generation, the process fails with the following error:
Enhance prompt failed: AI_APICallError: reasoning_effort "minimal" is not supported by model gpt-5-mini; supported values: [low medium high]
This appears to be due to a hardcoded reasoning_effort: "minimal" parameter in the system prompt enhancement and title generation features. My specific model provider's version of GPT-5-Mini only supports "low", "medium", or "high", and does not accept "minimal".
Steps to Reproduce:
Configure Kilo Code to use a custom model provider that offers gpt-5-mini but only supports reasoning_effort values of low, medium, or high.
Set this gpt-5-mini model as the one used for the "Enhance Prompt" feature.
Use the "Enhance Prompt" feature or try to generate a conversation title.
Observe the error message in the UI or logs.
Expected Behavior:
Kilo Code should allow configuration of the reasoning_effort value used for these background tasks. Either:
It should respect the model's advertised capabilities and not send "minimal" if unsupported.
Or, ideally, provide a setting where users can manually specify the desired reasoning effort (e.g., "low") for these features.
Actual Behavior:
The reasoning_effort parameter is set to "minimal" internally, causing an API error and preventing the feature from working. The official OpenAI API documentation confirms gpt-5-mini supports minimal, low, medium, and high, but my provider's implementation does not, highlighting the need for a configurable fallback.
Plugins
No response
Kilo version
7.2.10
Steps to reproduce
- Connect Github Copilot provider.
- Set small model to gpt-5-mini by Github Copilot
- Try to click the prompt enhanced button.
Screenshot and/or share link
No response
Operating System
No response
Terminal
No response
Description
Description:
When using a GPT-5-Mini model (via a custom provider) as the model for prompt enhancement or title generation, the process fails with the following error:
Enhance prompt failed: AI_APICallError: reasoning_effort "minimal" is not supported by model gpt-5-mini; supported values: [low medium high]
This appears to be due to a hardcoded reasoning_effort: "minimal" parameter in the system prompt enhancement and title generation features. My specific model provider's version of GPT-5-Mini only supports "low", "medium", or "high", and does not accept "minimal".
Steps to Reproduce:
Configure Kilo Code to use a custom model provider that offers gpt-5-mini but only supports reasoning_effort values of low, medium, or high.
Set this gpt-5-mini model as the one used for the "Enhance Prompt" feature.
Use the "Enhance Prompt" feature or try to generate a conversation title.
Observe the error message in the UI or logs.
Expected Behavior:
Kilo Code should allow configuration of the reasoning_effort value used for these background tasks. Either:
It should respect the model's advertised capabilities and not send "minimal" if unsupported.
Or, ideally, provide a setting where users can manually specify the desired reasoning effort (e.g., "low") for these features.
Actual Behavior:
The reasoning_effort parameter is set to "minimal" internally, causing an API error and preventing the feature from working. The official OpenAI API documentation confirms gpt-5-mini supports minimal, low, medium, and high, but my provider's implementation does not, highlighting the need for a configurable fallback.
Plugins
No response
Kilo version
7.2.10
Steps to reproduce
Screenshot and/or share link
No response
Operating System
No response
Terminal
No response