Shotcraft
Voice and speech

gpt-realtime-2.1

OpenAI's speech-to-speech model for live conversation, taking audio in and producing audio out with tool use.

A speech model by OpenAI · read 2026-08-14 · sourcing medium

Weights
Not released
API
Yes
Commercial use
Vendor states commercial use is permitted

A mini variant trades quality for roughly a third of the cost. A separate gpt-realtime-translate model does streaming speech-to-speech translation.

Commercial use — Vendor states commercial use is permitted

OpenAI API terms; commercial use permitted.

The licence notes recorded during research were read for a commercial-use position and did not state one. Read the licence before you rely on any of this. Nothing on this page is legal advice, and this site does not tell you a model is cleared for commercial use.

Specification

Vendor figures, read 2026-08-14Read from OpenAI's own documentation on 2026-08-14. Vendors change these between model versions without announcing it, so check theirs before you commit a shot to a number. Sourcing for this entry is mixed. Some fields came from pages we retrieved, others could not be confirmed and are left blank. Check the vendor's current documentation

What it takes in
audio, text
What it makes
audio, text
Controls
streaming-low-latency, style-prompting, emotion-control
Weights
Not released
API
Yes
Pricing
Audio tokens: gpt-realtime-2.1 $32.00/1M input, $0.40/1M cached input, $64.00/1M output. gpt-realtime-2.1-mini $10.00/1M input, $0.30/1M cached, $20.00/1M output. gpt-realtime-translate $0.034/minute. gpt-audio-1.5 $32.00/$64.00 per 1M; gpt-audio-mini $10.00/$20.00 per 1M.
Released or updated
not established
Languages
not established

Strengths

  • True speech-to-speech, preserving prosody rather than round-tripping through text
  • Very cheap cached input pricing ($0.40/1M) rewards long system prompts
  • Tool use and reasoning in the same model
  • Dedicated streaming translation variant

Weaknesses

  • Audio output at $64/1M tokens is expensive for long sessions
  • No voice cloning
  • gpt-realtime mini (the earlier one) is already deprecated - fast churn
  • Little direct application to non-interactive video production

Best for

Where the sources disagree

Recorded rather than resolved. Picking the tidier number would hide the fact that the vendor's own pages do not agree.

What to understand before using it

Explainers naming the mechanisms this model's own documentation mentions.

MCP servers that front OpenAI

Servers in the media MCP directory whose own description or homepage names this vendor. Absence of a server here is not evidence that none exists: most publish no tool list at all.

See it from the vendor

Shotcraft hosts no sample output. Judging a model means watching its own showreel with your own eyes, and a clip re-hosted here would be a clip we chose.

Nearby in the directory

gpt-4o-mini-tts

OpenAI's current text-to-speech model, steerable with natural-language instructions about how a line should be delivered. It replaces the older tts-1

Voice and speech

OpenAI transcription (gpt-4o-transcribe, gpt-transcribe, Whisper)

A family of speech-to-text models spanning batch and realtime use: Whisper and gpt-4o-transcribe for files, gpt-4o-transcribe-diarize for speaker-sepa

Voice and speech

Azure AI Speech text-to-speech (Neural, Neural HD, MAI voices)

Microsoft's enterprise speech synthesis service, with 200+ prebuilt neural voices across 80+ locales plus tiers for HD voices, multilingual voices, mu

Voice and speech

Sources

Read on 2026-08-14. A source that has changed since then will not be reflected here until the next pass.