Back to models
OpenAI logo

OpenAI

gpt-realtime-mini

A cost-efficient realtime model for voice and streaming interactions over WebRTC, WebSocket, or SIP.

Overall score
84
realtimevoicestreaming
Context window
32K tokens
Speed
Fast
Input pricing
$0.60 / 1M text input tokens
Output pricing
$2.40 / 1M text output tokens

Score breakdown

Capability79
Use-case fit88
Cost efficiency89
Speed98
Reliability84
Agent readiness87
Ecosystem90

Scores combine benchmark signals, product experience, and editorial weighting. Use them as a practical guide, not an absolute truth claim.

Best for

Agent automation

Works with

Voice agentsWebRTC appsTelephony

Modalities

textaudioimage

Sources & trust

Officially verified core fields
Official linkSummaryDescriptionModalitiesModality profileContext windowMax outputPricingPricing page

Editorial fields such as shortlist guidance, strengths, caveats, and scoring remain clearly separated from official provider data.

Last verified: Apr 9, 2026

Strengths

  • Low-latency audio and text interaction is the main strength.
  • Useful when UX speed matters more than deep long-form reasoning.

Things to watch

  • Smaller context and output limits than general frontier text models.
  • Not the right default for research-heavy or long-document tasks.

Best for