OpenAI speech API pricing and models

Price (provider's unit)

$0.0045 / minute

“| gpt-transcribe | Transcription | - | - | $0.0045 / minute |”

Converted $/hour (our arithmetic)

• $0.27converted from $0.0045 per minute

Languages

Not stated

Streaming

Yes

“Set stream=true with gpt-transcribe. The Transcriptions API returns transcript events as the model transcribes each part of the recording.”

Max audio

25 MB per file

“Files can be up to 25 MB. Supported input formats are mp3, mp4, mpeg, mpga, m4a, wav, and webm.”

Status

current (no preview label)

“Start with gpt-transcribe. This is the recommended model for transcribing recorded speech in its original language.”

No language count is stated for gpt-transcribe; the guide gives a count only for Whisper (98).

Price (provider's unit)

$0.017 / minute

“| gpt-live-transcribe | Live transcription | - | - | $0.017 / minute |”

Converted $/hour (our arithmetic)

• $1.02converted from $0.017 per minute

Languages

Not stated

Streaming

Yes

“GPT Live Transcribe is a streaming speech-to-text model for applications that need low-latency transcript deltas from live audio.”

Max audio

Not stated

Status

current (no preview label)

“| Feb 26, 2027 | whisper-1 | gpt-live-transcribe or gpt-transcribe |”

Named by OpenAI as the replacement for whisper-1 and the gpt-4o transcribe family. Endpoint: v1/realtime/transcription_sessions.

Price (provider's unit)

$0.017 / minute

“| gpt-realtime-whisper | Live transcription | - | - | $0.017 / minute |”

Converted $/hour (our arithmetic)

• $1.02converted from $0.017 per minute

Languages

Not stated

Streaming

Yes

“GPT-Realtime-Whisper is a streaming speech-to-text model for applications that need low-latency transcript deltas from live audio.”

Max audio

Not stated

Status

current (no preview label)

“GPT-Realtime-Whisper is priced by audio duration rather than text tokens.”

Model page lists a 16,000 context window and 2,000 max output tokens.

Price (provider's unit)

$0.003 / minute ($1.25 input / $5.00 output per 1M tokens)

“| gpt-4o-mini-transcribe | Transcription | $1.25 | $5.00 | $0.003 / minute |”

Converted $/hour (our arithmetic)

• $0.18converted from $0.003 per minute

Languages

Not stated

Streaming

Yes

“Existing gpt-4o-transcribe, gpt-4o-mini-transcribe, and gpt-4o-transcribe-diarize integrations also support file streaming. whisper-1 doesn't.”

Max audio

25 MB per file

“The Transcriptions API accepts files up to 25 MB.”

Status

deprecated (shutdown 2027-02-26)

“| Feb 26, 2027 | gpt-4o-mini-transcribe | gpt-live-transcribe or gpt-transcribe |”

Same shutdown date (2027-02-26) applies to whisper-1, gpt-4o-transcribe and gpt-4o-transcribe-diarize. The pricing page lists gpt-4o-transcribe at $0.006 / minute; whisper-1 and gpt-4o-transcribe-diarize no longer appear on it.

Price (provider's unit)

$0.60 per 1M text input tokens; $12.00 per 1M audio output tokens

“Price $0.6•$12 Input•Output”

Note: Model page units: 'Text tokens Per 1M tokens Input $0.60' and 'Audio tokens Per 1M tokens Output $12.00'.

Languages

follows Whisper language support

“The TTS model generally follows the Whisper model in terms of language support.”

Streaming

Yes

“The Speech API provides support for realtime audio streaming using chunk transfer encoding.”

Limit per request

2,000 input tokens

“The maximum number of input tokens is 2000.”

Status

deprecated (shutdown 2027-01-06)

“| Jan 6, 2027 | gpt-4o-mini-tts-2025-12-15 | gpt-realtime-2.1-mini |”

Every OpenAI TTS model is on the shutdown list: tts-1, tts-1-hd and both gpt-4o-mini-tts snapshots shut down Jan 6, 2027, with gpt-realtime-2.1-mini as the recommended replacement. tts-1 and tts-1-hd are not listed on the current pricing page. The TTS guide still calls gpt-4o-mini-tts 'our newest and most reliable text-to-speech model'.

Price (provider's unit)

$32.00 per 1M audio input tokens; $0.40 cached; $64.00 per 1M audio output tokens

“| gpt-realtime-2.1 | Audio | $32.00 | $0.40 | $64.00 |”

Languages

Not stated

Streaming

Not stated

Limits

Not stated

Status

current (no preview label)

“| Jan 20, 2027 | gpt-realtime | gpt-realtime-2.1 |”

Text tokens: $4.00 in / $0.40 cached / $24.00 out per 1M. 128,000 context window. Streaming left null: the model page lists the v1/realtime endpoint but its Features block reads 'Streaming Not supported' (likely the Responses streaming flag); no sentence on the page states realtime audio streaming.

Price (provider's unit)

$10.00 per 1M audio input tokens; $0.30 cached; $20.00 per 1M audio output tokens

“| gpt-realtime-2.1-mini | Audio | $10.00 | $0.30 | $20.00 |”

Languages

Not stated

Streaming

Yes

“It supports audio and text inputs over WebRTC, WebSocket, or SIP connections and improves alphanumeric recognition over GPT-Realtime-2.”

Note: The model page's Features block separately reads 'Streaming Not supported'.

Limits

Not stated

Status

current (no preview label)

“| Jan 6, 2027 | tts-1 | gpt-realtime-2.1-mini |”

Named by OpenAI as the replacement for all TTS models.

Price (provider's unit)

$0.05 per minute

“Voice sessions cost $0.05 per minute, billed per second. Backend model and tool usage is billed separately.”

Languages

Not stated

Streaming

Yes

“Streaming Supported Function calling Supported”

Limits

Not stated

Status

current (no preview label)

“| gpt-live-1 | $0.05 |”

Per-minute session price excludes backend model and tool usage.