Feature Request: Better Progress Indicators for the Responses API

I’m developing a personal project called The Arkivist, a collection cataloging application that uses the OpenAI API to identify objects from photographs and extract information into inventory fields.

The AI does a remarkably good job. I’ve been very impressed with the results.

One thing I’ve struggled with, however, is giving users meaningful feedback while the AI is working.

I can tell when a request starts, I can receive streamed response data, and I can tell when the request is finished. But I have no reliable way of knowing how far along the actual process is.

Is it analyzing the photographs? Identifying the object? Extracting text? Preparing its response? Is it almost finished, or does it have another 15 seconds to go?

I don’t know.

I’ve recently built a visual progress indicator to make the experience a little better. It moves through four stages: submitting photographs, identifying the item, preparing the data, and completion.

It looks nice, but the percentages are essentially milestones that I’ve programmed myself. They don’t reflect the actual progress of the AI.

What I’d really love to see is some way for the API to report genuine processing progress.

Even something relatively simple would help:

  • Notifications when major processing stages begin and end.

  • An approximate completion percentage, when possible.

  • Perhaps an estimated remaining time.

I completely understand that AI processing isn’t necessarily linear, and that an accurate percentage might not always be possible. I’m not looking for access to internal reasoning or chain-of-thought. I just want to give people using my application a better idea of what’s happening while they wait.

The Responses API streaming capabilities are already very useful. Some additional progress information would make them even more useful.

Thanks for considering the suggestion!