On November 20, 2024, OpenAI announced significant updates to its platform for developers, enhancing the capabilities available for testing and evaluation.

Among the new features is the Evaluations tool, which allows developers to input multiple user scenarios to assess a model's prompt performance against a benchmark.

However, it remains unclear whether the Evaluations feature will be accessible through the API, which could limit its integration into continuous development processes.

Additionally, developers can now utilize a microphone to dictate inputs and listen to audio outputs, all while viewing the transcribed text generated by the audio model.

This voice output feature is particularly beneficial for users who do not have access to ChatGPT Plus, expanding the tool's usability.

Notably, a similar feature has already been available on Anthropic Workbench, allowing developers to create specific test cases for their prompts.