What is your goal?
I’m sending a small MP3 audio file (a few MB) to OpenAI’s Whisper transcription endpoint (https://api.openai.com/v1/audio/transcriptions) using the HTTP module with multipart/form-data, so I can get back subtitle (.srt) content for a video production pipeline.
What is the problem & what have you tried?
The HTTP module returns a 400 error saying “Part exceeded maximum size of 1024KB” even though my file is only a few MB, well under Make’s documented Pro plan limit (250MB).
Setup: HTTP module, POST, multipart/form-data body with fields: model (whisper-1), file (binary, mapped from a previous HTTP GET module’s binary output), language (ja), response_format (srt), prompt (text).
I initially got a different error (“Unrecognized file format”) which I fixed by explicitly adding a filename to the file field mapping. After that fix, this new 1024KB size error appeared instead.
I’ve confirmed my account is on the Pro plan, which per Make’s documentation should allow payloads up to 250MB, so this 1024KB limit doesn’t match. I suspect this might be a hardcoded internal limit on individual multipart “parts” specifically, separate from the overall account payload cap — but I haven’t found documentation confirming this.
I’ve tried:
- Re-checking the file field is set to binary/file type, not text
- Confirming the source file itself is small (a few MB)
- Verifying the filename fix from the previous error is still in place
I have not found a way to see or adjust an internal per-part size setting, if one exists.
Error messages or input/output bundles
{
“error”: {
“message”: “400: Part exceeded maximum size of 1024KB.”,
“type”: “server_error”,
“param”: null,
“code”: null
}
}
Create public scenario page
https://eu1.make.com/public/shared-scenario/2vV4kHkuUHF/02-audio-processed-generate-subtitles
Screenshots (scenario flow, module settings, errors)
