According to ITmedia AI+, ChatGPT now accepts audio file uploads for paid users, with transcription, summaries and questions about the content. The file size limit is 512 MB.
What was announced
In an article dated October 9, 2026, ITmedia AI+ reports that ChatGPT can now read audio files. It is aimed at paid users, who can transcribe audio, summarize it and ask questions about what was said.
The article says supported formats include WAV, MP3, OGG, FLAC, AAC and M4A, and that audio-only WebM and MP4 files also work. The file size limit is 512 MB. Step-by-step instructions are reportedly given in OpenAI's help page, 'Uploading files and audio to ChatGPT'.
Points to watch
According to the article, long recordings may be split or the processing may time out. Transcripts can contain errors in the content or in identifying speakers, and accuracy may vary by language.
The article does not go into detail on which plans are covered or on accuracy for Japanese. Check OpenAI's official help pages before relying on the feature.
TAKEAWAY
What this means for your project
This may suit drafting meeting notes from recorded meetings or calls. Before uploading recordings with customer names or contact details, confirm consent and your internal data-handling rules. Transcripts can contain errors, so check amounts, dates and names against the audio. Start with low-sensitivity recordings, such as internal meetings, to judge accuracy.