Hello,
This is GPT Prompt Tester.
In the v0.9.5 update, the Eval feature has been expanded with the addition of a new Response Quality analysis capability.
We have also improved Microsoft Store subscription stability and refined parts of the internal architecture.
Here are the main updates in this release.
✨ Key Updates
1️⃣ Eval Feature Expansion — Response Quality Analysis
This update introduces a new Response Quality analysis feature that evaluates the quality of GPT-generated responses.
Previously, the Eval feature focused mainly on prompt design quality analysis.
With this update, it can now also evaluate the quality of the responses generated by the model itself.
The Response Quality analysis evaluates responses based on the following criteria:
Correctness
The factual accuracy of the response and the likelihood of errors.
Alignment
How well the response matches the user’s request and prompt intent.
Reasoning
The logical structure and clarity of the explanation.
Effectiveness
How clearly and effectively the response communicates its message.
Instruction Adherence
How well the response follows the prompt rules and requirements.
Each category includes a score along with analysis comments, allowing users to easily identify strengths and improvement points in the response.
In addition, each evaluation includes detailed analysis sections:
Observations — Key characteristics observed in the response
Risks — Potential issues or weaknesses
Suggestions — Recommendations for improvement
This means the feature provides not just scores, but a practical analysis report that helps improve response quality.
With this addition, GPT Prompt Tester now naturally supports the following prompt improvement workflow:
Prompt Analysis → Response Quality → Improve
This enables a complete experiment and improvement cycle within the app—from prompt design analysis to response evaluation and refinement.
2️⃣ Overall Stability and Internal Improvements
This update also includes improvements to the app’s internal structure.
Key improvements include:
- Improved Microsoft Store subscription stability
- Improved image and thumbnail caching structure
- Internal module refactoring and performance improvements
These changes improve the overall stability and consistency of the application.
Going forward, we will continue improving GPT Prompt Tester so it can serve as a practical tool for prompt experimentation and GPT-powered workflows.
If you encounter any issues or have suggestions, please feel free to let us know.
Your feedback greatly helps us improve the service.
Thank you for using GPT Prompt Tester. 😊
