OpenAI Whisper API
OpenAI's hosted Whisper API — gpt-4o-transcribe is the newest model, gpt-4o-mini-transcribe is the cheaper sibling, whisper-1 is the legacy reference. One REST surface.
Drop your audio. Transcript in seconds. First transcript free, then $2 a file or $8 = 1,000 min
OpenAI exposes hosted speech-to-text on three models behind a single REST endpoint, POST https://api.openai.com/v1/audio/transcriptions: gpt-4o-transcribe and gpt-4o-mini-transcribe (newer GPT-4o-class transcribers, faster and tuned for noisier audio) and whisper-1 (the legacy reference model running Whisper large-v2). The same surface also serves audio-to-English at /v1/audio/translations, and the parallel Realtime API (gpt-4o-realtime-preview) carries streaming STT over WebSocket / WebRTC for live agents.
Best for teams already paying for OpenAI infra, anyone mixing transcription with chat / function-calling in the same SDK, or workflows where vendor consolidation beats per-minute price. Current per-minute rates: gpt-4o-transcribe at $0.006/min, gpt-4o-mini-transcribe at $0.003/min, whisper-1 at $0.006/min, billed to the nearest second. Hard limits: 25 MB per request, formats mp3 / mp4 / mpeg / mpga / m4a / wav / webm, no built-in diarization, no batch discount.
What it is
OpenAI's hosted Whisper API is the easiest way to get Whisper-grade transcription without running infrastructure. $0.006 per minute, pay-as-you-go. No diarization, no streaming — if you need those, pick a different endpoint or self-host whisperX. Last price check: 2026-04-20.
Watch out for: 25 MB file size limit; no diarization; no batch discount; latency dominated by upload for large files.
Install / use
Where the Whisper API fits · 6 use-cases
OpenAI's audio surface is small but covers the common shapes: subtitles, translation, long-form chunking, real-time, telephony, and multilingual. Each card points at the matching section on platform.openai.com — pick the closest one and copy a recipe below.
Pass response_format=srt or response_format=vtt to /v1/audio/transcriptions and the API returns a subtitle file directly — no client-side stitching. For finer timing pass response_format=verbose_json with timestamp_granularities=['segment','word'] to get per-segment and per-word offsets on whisper-1 or gpt-4o-transcribe.
verbose_json + timestamp_granularities for word-level offsets
The sibling endpoint /v1/audio/translations transcribes non-English audio directly into English text in one call — useful when downstream tooling is English-only. Currently supported on whisper-1; for other source/target pairs run /transcriptions then a chat model.
Translation task is English-only output
Files over 25 MB must be split client-side before upload. The OpenAI cookbook pattern uses pydub to slice on silence boundaries, transcribe each chunk in parallel, then concatenate. Alternative: compress to 16 kHz mono opus / m4a to fit a 60-90 min episode under 25 MB.
Hard 25 MB cap per request — no server-side chunking
For sub-second streaming STT use the Realtime API (gpt-4o-realtime-preview), not /audio/transcriptions — the REST endpoint is request/response only. Realtime carries bidirectional audio over WebSocket or WebRTC and is the path for voice agents, live captions, and IVR replacements.
Separate SKU and endpoint from /audio/transcriptions
gpt-4o-mini-transcribe at $0.003/min is the budget pick for high-volume call recordings and voicemail batches where every cent matters and word-level timestamps are optional. No diarization in the response — pair with a downstream speaker-attribution model or a tool that bundles it.
Half the per-minute cost of gpt-4o-transcribe and whisper-1
All three models auto-detect the input language; pass an ISO-639-1 language hint (e.g. language='ja') to skip detection and improve accuracy on short clips. Quality varies by language — large-v2 weights underneath whisper-1 are the same set the open-source Whisper community benchmarks against.
Optional language hint sharpens short-clip detection
Quickstart · pick a runtime
Three minimal calls to /v1/audio/transcriptions with gpt-4o-transcribe. Export your key as OPENAI_API_KEY first — get one from the OpenAI dashboard. Never hard-code the key in source.
Official openai Python SDK · transcribe a local file with gpt-4o-transcribe.
# pip install --upgrade openai
import os
from openai import OpenAI
client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])
with open("audio.mp3", "rb") as f:
resp = client.audio.transcriptions.create(
model="gpt-4o-transcribe",
file=f,
# response_format="verbose_json",
# timestamp_granularities=["segment", "word"], # whisper-1 + gpt-4o-transcribe
# language="en", # ISO-639-1 hint
)
print(resp.text)Official openai Node / TypeScript SDK · same call from Node 18+.
// npm install openai
import fs from "node:fs";
import OpenAI from "openai";
const client = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });
const resp = await client.audio.transcriptions.create({
model: "gpt-4o-transcribe",
file: fs.createReadStream("audio.mp3"),
// response_format: "verbose_json",
// timestamp_granularities: ["segment", "word"],
// language: "en",
});
console.log(resp.text);Plain HTTPS POST to /v1/audio/transcriptions · works from shell, CI, and edge runtimes.
# Bearer auth via $OPENAI_API_KEY · multipart upload
curl https://api.openai.com/v1/audio/transcriptions \
-H "Authorization: Bearer $OPENAI_API_KEY" \
-H "Content-Type: multipart/form-data" \
-F file="@audio.mp3" \
-F model="gpt-4o-transcribe"
# Subtitle output: add -F response_format="srt"
# Word timestamps (whisper-1 / gpt-4o-transcribe):
# -F response_format="verbose_json" \
# -F "timestamp_granularities[]=word" -F "timestamp_granularities[]=segment"What people actually do with OpenAI Whisper API-style transcription
The tool is the means. These are the jobs — each one priced at published rates, each one wired up on its own page.
Features
| Speaker diarization | No |
| Word-level timestamps | Yes |
| Streaming / real-time | No |
| Languages supported | 99 |
| HIPAA eligible | No |
Links
- platform.openai.com/docs/guides/speech-to-text ↗Canonical Speech-to-text guide — model list, response_format options, timestamp_granularities, chunking patterns, language hints.
- API reference · createTranscription ↗Full parameter surface for /v1/audio/transcriptions — file, model, prompt, response_format, temperature, language, timestamp_granularities.
- openai.com/api/pricing ↗Live per-minute rates for gpt-4o-transcribe ($0.006/min), gpt-4o-mini-transcribe ($0.003/min), and whisper-1 ($0.006/min).
- status.openai.com ↗Live status for the API, including Audio endpoints — subscribe via email or RSS.
- openai/openai-python ↗Official Python SDK — v2.x line, audio.transcriptions.create + audio.translations.create, sync + async clients.
- openai/openai-node ↗Official JavaScript / TypeScript SDK — Node 18+, browser, edge runtimes; same audio.transcriptions surface.
- openai/openai-cookbook ↗Official examples repo — see the examples/ folder for audio-chunking patterns (pydub silence-split) and verbose_json post-processing.
- platform.openai.com/docs/guides/realtime ↗Realtime API guide — gpt-4o-realtime-preview over WebSocket / WebRTC for live STT, voice agents, and barge-in conversation loops.
OpenAI Whisper API vs Whipscribe
| Feature | OpenAI Whisper API | Whipscribe |
|---|---|---|
| Category | Transcription APIs | Transcription APIs |
| Pricing | Not verified | $8–$24 one-time packs (credits never expire) · $2 single unlock · free instant preview |
| Speaker diarization | Not verified | Yes |
| Word timestamps | Not verified | Yes |
| Streaming | Not verified | No |
| Languages | 99 | 99 |
| Platforms | API | Web, API, MCP |
Where this category is heading
From the vendor changelogs we track weekly — what changed in August 2026, and what it means if you are choosing now.
AssemblyAI moved summarisation onto an LLM this month; every vendor is racing to return action items, quotes and topics with the text rather than as an add-on.
Whipscribe today Every Whipscribe job already returns an insights payload — summary, key quotes, topics and speakers — from the same job id, at no extra charge.
Deepgram shipped self-hosted container images in August — the market is moving toward audio that stays inside a boundary the customer controls, because teams with customer calls or unreleased material are refusing shared model endpoints.
Whipscribe today Whipscribe runs on our own GPUs in a private, secured cloud. Audio is never forwarded to OpenAI or any third-party model.
The fastest-growing way to use a transcription API is not a form — it is Claude, Cursor or a workflow runner calling it mid-task through MCP.
Whipscribe today Whipscribe ships an MCP server: transcribe, search and summarise from an assistant without wiring anything.
AssemblyAI's 1.0 SDK unified async, realtime and sync; Deepgram's CLI went to 0.3. The unit of work is becoming the folder or the bucket, not the file.
Whipscribe today Submit with an Idempotency-Key and a batch_id, poll by job, retry safely. The S3 connector runs a whole prefix in one grant.
Deepgram added Afrikaans, Georgian and Armenian and improved a dozen more this month. Coverage is widening while quality still clusters around English and the large European languages.
Whipscribe today 99+ languages auto-detected. Ask for a language explicitly when you know it — auto-detect on a short or noisy clip is the most common cause of a wrong-language transcript.
Source: Deepgram and AssemblyAI changelogs, scanned 2026-08-24.
Alternatives to OpenAI Whisper API
Frequently asked about OpenAI Whisper API
What is OpenAI Whisper API?
OpenAI exposes hosted speech-to-text on three models behind a single REST endpoint, POST https://api.openai.com/v1/audio/transcriptions: gpt-4o-transcribe and gpt-4o-mini-transcribe (newer GPT-4o-class transcribers, faster and tuned for noisier audio) and whisper-1 (the legacy reference model running Whisper large-v2). The same surface also serves audio-to-English at /v1/audio/translations, and the parallel Realtime API (gpt-4o-realtime-preview) carries streaming STT over WebSocket / WebRTC for live agents.
How much does OpenAI Whisper API cost?
OpenAI Whisper API is a paid product — published pricing: $0.006/min. Pricing changes; verify on the vendor's own page before budgeting.
What platforms does OpenAI Whisper API support?
OpenAI Whisper API is an API — you call it from whatever you build, on any platform with an HTTP client.
How do I get started with OpenAI Whisper API?
OpenAI Whisper API is used through its API: get a key from the vendor, then call it from your own code. There is no desktop app to install.
How many languages does OpenAI Whisper API support?
OpenAI Whisper API lists 99 languages. That is the Whisper-family multilingual set, so quality varies by language — English and the large European languages are strongest.
What are the limitations of OpenAI Whisper API?
25 MB file size limit; no diarization; no batch discount; latency dominated by upload for large files.
Who is OpenAI Whisper API best for?
Teams already on OpenAI's stack who want Whisper without operating a GPU.
What should I know before choosing OpenAI Whisper API?
Best for teams already paying for OpenAI infra, anyone mixing transcription with chat / function-calling in the same SDK, or workflows where vendor consolidation beats per-minute price. Current per-minute rates: gpt-4o-transcribe at $0.006/min, gpt-4o-mini-transcribe at $0.003/min, whisper-1 at $0.006/min, billed to the nearest second. Hard limits: 25 MB per request, formats mp3 / mp4 / mpeg / mpga / m4a / wav / webm, no built-in diarization, no batch discount.
Is OpenAI Whisper API open source?
No. OpenAI Whisper API is a proprietary API. If you need source you can read and run yourself, the open-source tools in this directory are the place to look.
What kind of tool is OpenAI Whisper API?
In this directory OpenAI Whisper API is filed under commercial api as a API.
What are the alternatives to OpenAI Whisper API?
There is a side-by-side page at /tools/openai-whisper-api-alternatives comparing OpenAI Whisper API with the closest tools in the same category on price, platform and features.
Are there setup recipes for OpenAI Whisper API?
Yes — this page carries 3 tested setups: Python SDK · openai-python v2.x; Node SDK · openai-node v4+; cURL · multipart/form-data.
Can I automate OpenAI Whisper API-style transcription without running it myself?
Yes. If what you want is transcripts rather than the tool itself, Whipscribe does the same job as a hosted API: submit a file or URL, poll a job id, pull the result as txt, json, srt, vtt or docx. $0.008 a minute — about $0.48 an audio hour — bought as credits that never expire. The automation recipes on this site show it wired to Drive folders, Zoom recordings, S3 buckets and no-code platforms.
How is Whipscribe different from OpenAI Whisper API?
OpenAI Whisper API is another vendor's product; Whipscribe is a per-minute transcription service. We run the models on our own GPUs in a private cloud — audio is never forwarded to OpenAI or any third-party model — and every job returns five formats plus an AI summary, quotes and topics. Your first transcript in the web app is free at any length, no card.
What does transcription cost if I use Whipscribe instead of OpenAI Whisper API?
$0.008 a minute — about $0.48 an audio hour — bought as credits that never expire. There is no subscription and no monthly minimum: a 3-hour recording is about $1.44, a thousand-hour archive is about $480 once. Your first transcript in the web app is free at any length, no card, so you can check accuracy on your own audio before paying anything.
Whipscribe is a managed faster-whisper + whisperX service. If you want transcripts without running infrastructure, paste a URL or drop a file in the form below — you'll have a transcript in seconds.
Explore
All transcription tools · Audio technology hub · Transcribe any platform · Audio & video formats · How-to guides · Glossary · Playbooks · Apps · Broadcast & radio · Podcast transcripts · Use cases · Blog · Transcription API · Integrations · Automations · For your industry