GPT-4o Transcribe is OpenAI’s speech-to-text model powered by GPT-4o. Its input is audio with optional text guidance, and its output is recognized text. It is separate from the GPT-4o Mini Transcribe edition and from GPT-4o Transcribe Diarize, the specialized speaker-labeling model. Use this profile to prepare a faithful textual record of spoken material, with the source recording available for review and later editorial work.
OpenAI’s transcription guide documents uploads up to 25 MB. The current endpoint reference accepts FLAC, MP3, MP4, MPEG, MPGA, M4A, OGG, WAV and WebM and recommends file metadata that identifies the format, including an extension-bearing filename and an appropriate content type. For a larger recording, compression or ordered pieces of 25 MB or less follow the producer’s longer-input guidance. Avoid a split in the middle of a sentence, and retain the position of each piece in the full recording.
For an interview, associate each input file with the recording and the passage it covers. For a technical presentation, prepare the relevant product names, acronyms and domain vocabulary as context, then check those spellings against what the speaker actually said. For a recorded explanation with numerical values, review the transcript alongside the source before turning it into a written procedure. The desired first artifact is recognized speech in text form; a condensed report or translated document can be prepared from that text as a separate step.
The producer’s guide explicitly includes GPT-4o Transcribe in its prompting support. A prompt can supply correct spellings, context from a prior segment and preferences for punctuation, capitalization, filler words or a writing system. The reference says it should match the recording’s language. A short, relevant description of the topic and vocabulary gives the transcription useful context without treating unspoken terms as required output.
This profile supports the singular language hint, documented in ISO 639-1 form. Supplying the actual input language can aid recognition. GPT Transcribe’s newer keywords and plural languages fields belong to that separately named model and should not be assumed for GPT-4o Transcribe. Keep the language setting tied to the speech you are transcribing, and preserve the original wording before any later translation or stylistic rewrite.
The current endpoint reference documents JSON as GPT-4o Transcribe’s supported response format. The result contains recognized text, and optional token log probabilities can be requested with that JSON format. The probability data describe the model’s selection of tokens. They can inform which passages deserve inspection, but an audio comparison establishes whether the transcript reproduces a difficult name, number or phrase.
The standard profile’s JSON result does not supply native subtitle formats, verbose word timestamps or speaker assignments. OpenAI documents other model-and-format choices when those outputs are needed; GPT-4o Transcribe Diarize is the distinct speaker-labeling edition. For a captioning task, retain a checked transcript for a suitable timing step. For a conversation, do not invent speaker names merely because a transcript paragraph seems to change topic or voice.
OpenAI’s speech-to-text guide says existing GPT-4o Transcribe integrations support file streaming. The application supplies the recording first, then receives transcript text as that recording is processed and retains the completed result. This streams the transcript of a completed recording rather than accepting microphone audio as it arrives. The exact model card also lists transcription and Realtime endpoint families; model-specific session behavior must remain associated with this profile rather than borrowed from a newer continuously streaming model.
The current GPT-4o Transcribe specification lists a 16,000-token context window, a maximum output of 2,000 tokens and a knowledge cutoff of June 1, 2024. Those token limits are not a fixed audio duration and do not replace the file-transcription preparation rules. For an extended recording, keep the source order and relevant preceding context, then verify that the completed text reaches the end of each intended segment.
A useful final transcript can become searchable interview material, a source for technical documentation or the wording for a later caption workflow. Review uncertain vocabulary, quantities, punctuation, accents and overlapping speech against the recording. Keep recognized text distinct from later corrections or summaries, and retain the source link through identifiers in the surrounding workflow. The model supplies a text record; careful input preparation and source review determine whether that record is suitable for the next use.
", "aboutText": "GPT-4o Transcribe is OpenAI’s speech-to-text model powered by GPT-4o. Its input is audio with optional text guidance, and its output is recognized text. It is separate from the GPT-4o Mini Transcribe edition and from GPT-4o Transcribe Diarize, the specialized speaker-labeling model. Use this profile to prepare a faithful textual record of spoken material, with the source recording available for review and later editorial work.\n\n### File input and preparing long material\n\nOpenAI’s transcription guide documents uploads up to 25 MB. The current endpoint reference accepts FLAC, MP3, MP4, MPEG, MPGA, M4A, OGG, WAV and WebM and recommends file metadata that identifies the format, including an extension-bearing filename and an appropriate content type. For a larger recording, compression or ordered pieces of 25 MB or less follow the producer’s longer-input guidance. Avoid a split in the middle of a sentence, and retain the position of each piece in the full recording.\n\nFor an interview, associate each input file with the recording and the passage it covers. For a technical presentation, prepare the relevant product names, acronyms and domain vocabulary as context, then check those spellings against what the speaker actually said. For a recorded explanation with numerical values, review the transcript alongside the source before turning it into a written procedure. The desired first artifact is recognized speech in text form; a condensed report or translated document can be prepared from that text as a separate step.\n\n### Recording-specific prompts and language choice\n\nThe producer’s guide explicitly includes GPT-4o Transcribe in its prompting support. A prompt can supply correct spellings, context from a prior segment and preferences for punctuation, capitalization, filler words or a writing system. The reference says it should match the recording’s language. A short, relevant description of the topic and vocabulary gives the transcription useful context without treating unspoken terms as required output.\n\nThis profile supports the singular language hint, documented in ISO 639-1 form. Supplying the actual input language can aid recognition. GPT Transcribe’s newer keywords and plural languages fields belong to that separately named model and should not be assumed for GPT-4o Transcribe. Keep the language setting tied to the speech you are transcribing, and preserve the original wording before any later translation or stylistic rewrite.\n\n### Result format, probabilities and specialized outputs\n\nThe current endpoint reference documents JSON as GPT-4o Transcribe’s supported response format. The result contains recognized text, and optional token log probabilities can be requested with that JSON format. The probability data describe the model’s selection of tokens. They can inform which passages deserve inspection, but an audio comparison establishes whether the transcript reproduces a difficult name, number or phrase.\n\nThe standard profile’s JSON result does not supply native subtitle formats, verbose word timestamps or speaker assignments. OpenAI documents other model-and-format choices when those outputs are needed; GPT-4o Transcribe Diarize is the distinct speaker-labeling edition. For a captioning task, retain a checked transcript for a suitable timing step. For a conversation, do not invent speaker names merely because a transcript paragraph seems to change topic or voice.\n\n### Streamed results from a completed file\n\nOpenAI’s speech-to-text guide says existing GPT-4o Transcribe integrations support file streaming. The application supplies the recording first, then receives transcript text as that recording is processed and retains the completed result. This streams the transcript of a completed recording rather than accepting microphone audio as it arrives. The exact model card also lists transcription and Realtime endpoint families; model-specific session behavior must remain associated with this profile rather than borrowed from a newer continuously streaming model.\n\n### Model limits and checking the final artifact\n\nThe current GPT-4o Transcribe specification lists a 16,000-token context window, a maximum output of 2,000 tokens and a knowledge cutoff of June 1, 2024. Those token limits are not a fixed audio duration and do not replace the file-transcription preparation rules. For an extended recording, keep the source order and relevant preceding context, then verify that the completed text reaches the end of each intended segment.\n\nA useful final transcript can become searchable interview material, a source for technical documentation or the wording for a later caption workflow. Review uncertain vocabulary, quantities, punctuation, accents and overlapping speech against the recording. Keep recognized text distinct from later corrections or summaries, and retain the source link through identifiers in the surrounding workflow. The model supplies a text record; careful input preparation and source review determine whether that record is suitable for the next use.", "editorialOverlay": { "copy_fields": null, "copy_sha256": null, "copy_source": null, "html_sha256": "ddbf2668a5e461e01ecf1c13301918c96ebf5b325c7bed5126e37d149e668d86", "module_sha256": "ee9dc1c50b1e6a8fb7fa062030fcb4646b9aaaeb08be513b723a5a92a7290f06", "raw_model_unchanged": true, "schema": "ailoot-native-model-editorial-v1" }, "locale": "en", "methods": [ { "html": "openai/gpt-4o-transcribeGPT-4o Transcribe converts audio into text using OpenAI’s GPT-4o speech-recognition capabilities. It supports contextual prompts, an input-language hint, JSON transcripts, streamed file output and optional token log probabilities for reviewing uncertain wording.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
GPT-4o Transcribe is OpenAI’s speech-to-text model powered by GPT-4o. Its input is audio with optional text guidance, and its output is recognized text. It is separate from the GPT-4o Mini Transcribe edition and from GPT-4o Transcribe Diarize, the specialized speaker-labeling model. Use this profile to prepare a faithful textual record of spoken material, with the source recording available for review and later editorial work.
OpenAI’s transcription guide documents uploads up to 25 MB. The current endpoint reference accepts FLAC, MP3, MP4, MPEG, MPGA, M4A, OGG, WAV and WebM and recommends file metadata that identifies the format, including an extension-bearing filename and an appropriate content type. For a larger recording, compression or ordered pieces of 25 MB or less follow the producer’s longer-input guidance. Avoid a split in the middle of a sentence, and retain the position of each piece in the full recording.
For an interview, associate each input file with the recording and the passage it covers. For a technical presentation, prepare the relevant product names, acronyms and domain vocabulary as context, then check those spellings against what the speaker actually said. For a recorded explanation with numerical values, review the transcript alongside the source before turning it into a written procedure. The desired first artifact is recognized speech in text form; a condensed report or translated document can be prepared from that text as a separate step.
The producer’s guide explicitly includes GPT-4o Transcribe in its prompting support. A prompt can supply correct spellings, context from a prior segment and preferences for punctuation, capitalization, filler words or a writing system. The reference says it should match the recording’s language. A short, relevant description of the topic and vocabulary gives the transcription useful context without treating unspoken terms as required output.
This profile supports the singular language hint, documented in ISO 639-1 form. Supplying the actual input language can aid recognition. GPT Transcribe’s newer keywords and plural languages fields belong to that separately named model and should not be assumed for GPT-4o Transcribe. Keep the language setting tied to the speech you are transcribing, and preserve the original wording before any later translation or stylistic rewrite.
The current endpoint reference documents JSON as GPT-4o Transcribe’s supported response format. The result contains recognized text, and optional token log probabilities can be requested with that JSON format. The probability data describe the model’s selection of tokens. They can inform which passages deserve inspection, but an audio comparison establishes whether the transcript reproduces a difficult name, number or phrase.
The standard profile’s JSON result does not supply native subtitle formats, verbose word timestamps or speaker assignments. OpenAI documents other model-and-format choices when those outputs are needed; GPT-4o Transcribe Diarize is the distinct speaker-labeling edition. For a captioning task, retain a checked transcript for a suitable timing step. For a conversation, do not invent speaker names merely because a transcript paragraph seems to change topic or voice.
OpenAI’s speech-to-text guide says existing GPT-4o Transcribe integrations support file streaming. The application supplies the recording first, then receives transcript text as that recording is processed and retains the completed result. This streams the transcript of a completed recording rather than accepting microphone audio as it arrives. The exact model card also lists transcription and Realtime endpoint families; model-specific session behavior must remain associated with this profile rather than borrowed from a newer continuously streaming model.
The current GPT-4o Transcribe specification lists a 16,000-token context window, a maximum output of 2,000 tokens and a knowledge cutoff of June 1, 2024. Those token limits are not a fixed audio duration and do not replace the file-transcription preparation rules. For an extended recording, keep the source order and relevant preceding context, then verify that the completed text reaches the end of each intended segment.
A useful final transcript can become searchable interview material, a source for technical documentation or the wording for a later caption workflow. Review uncertain vocabulary, quantities, punctuation, accents and overlapping speech against the recording. Keep recognized text distinct from later corrections or summaries, and retain the source link through identifiers in the surrounding workflow. The model supplies a text record; careful input preparation and source review determine whether that record is suitable for the next use.
POST /v1/audio/transcriptionsaudiotext| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits openai/gpt-4o-transcribe | Default— |
Fieldinput_audio | Typeobject | RequirementRequired | Values & limits | Default— |
| Parameter | Contract |
|---|---|
Parametercontent_type | Contractapplication/json |
Parameterrequired | Contractmodel, input_audio |
Parametermaterial_types | Contractaudio |
Parametermaterial_conditions | ContractThe example reads your actual nonempty PCM WAV file, encodes the complete file bytes as base64 and sends input_audio.format as wav. |
Lifecycle: synchronous_result_then_owned_library
GET /v1/media/tasks/{task_id} · exact task_id from an existing owned media record
GET /v1/media/assets/{asset_id}/content
GET /v1/media/assets
GET /v1/media/assets/{asset_id}
Tariff references do not add request options. Supported parameters are listed for the selected method.
Match each charge to the conditions shown. Use one rate for each set of alternative configurations.
| Configuration | Rate schedule | Billing unit | Charge |
|---|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Billing unit— | Charge Input · Input tokens — / 1K tokens Output · Output tokens — / 1K tokens Input · Cache read — / 1K tokens Input · Cache write — / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
| Configurationinput token | Rate schedule— | Billing unit/ 1K tokens | Charge Input · input token $0.005 / 1K tokens |
| Configurationoutput token | Rate schedule— | Billing unit/ 1K tokens | Charge Output · output token $0.02 / 1K tokens |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
set -eu\n: "${AILOOT_API_KEY:?Set your Ailoot API key}"\n: "${AILOOT_AUDIO_FILE:?Set the path to your actual PCM WAV recording}"\nexport AILOOT_AUDIO_FILE\npython3 - <<'PY_REQUEST'\nimport base64, io, json, os, wave\nfrom pathlib import Path\naudio = Path(os.environ['AILOOT_AUDIO_FILE']).read_bytes()\nif not (audio[:4] == b'RIFF' and audio[8:12] == b'WAVE'):\n raise SystemExit('Supply your actual PCM WAV audio file')\nwith wave.open(io.BytesIO(audio), 'rb') as recording:\n if recording.getcomptype() != 'NONE' or recording.getnframes() <= 0:\n raise SystemExit('Supply a nonempty PCM WAV recording')\nrequest = {'model': 'openai/gpt-4o-transcribe', 'input_audio': {'data': base64.b64encode(audio).decode('ascii'), 'format': 'wav'}}\nPath('request.json').write_text(json.dumps(request, ensure_ascii=False), encoding='utf-8')\nPY_REQUEST\ncurl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/audio/transcriptions \\\n -H "Authorization: Bearer $AILOOT_API_KEY" -H 'Content-Type: application/json' \\\n --data-binary @request.json -D response-headers.txt -o response.bin\n# Retain the complete original response, headers and actual X-Client-Request-ID.\n# Read existing owned records for recovery; do not resubmit an uncertain operation.\nComplete original response body, status and forwarded headers. No universal JSON shape, text field or task envelope is added by this handler.
GET /v1/media/tasks/{task_id}GET /v1/media/assets/{asset_id}/contentGET /v1/media/assetsGET /v1/media/assets/{asset_id}Keep every complete returned response, file and content part.
Retain the complete response and headers, every actual output, original returned identifiers and all usage fields. Recover the original operation without submitting creation again.
Retain the actual X-Client-Request-ID and complete response. Read the original owned assets and their exact signed content URLs; never resubmit an uncertain creation.
| Field | Type |
|---|---|
Fieldtext (when present) | Type |
Fieldusage (when present) | Type |
| Response | Condition | Next step |
|---|---|---|
Response400 · invalid_request_error | ConditionThe JSON body or explicit canonical model ID is invalid. | Next stepRetain the response and correct the indicated request field. |
Response401 · authentication_error | ConditionThe Ailoot API key is invalid. | Next stepUse your authorized Ailoot API key. |
Response403 · permission_error | ConditionThe key group does not admit this method. | Next stepRetain the returned error details. |
Response404 · not_found_error | ConditionThe selected method has no admitted route. | Next stepRetain the returned error details. |
Response422 · price_unavailable | ConditionThe accepted tariff cannot obtain its required input quantity or bound. | Next stepRetain the response; use the input representation required by the accepted tariff. |
Response503 · | ConditionRequired storage, route, method price or reservation is unavailable. | Next stepKeep the complete original response. An uncertain reservation is not permission to resubmit. |
Response502 · server_error | ConditionThe original result, usage or settlement requires checking. | Next stepRetain the actual request ID and original response. Read the existing owned record; do not resubmit creation. |
Ailoot Ultra openai/gpt-4o-transcribe
Use the method paths shown above.