OpenAI Realtime API pricing (2026)
Short answer. OpenAI prices the Realtime API by tokens. gpt-realtime-2.1 costs $32.00 per 1M audio input tokens and $64.00 per 1M audio output tokens. gpt-realtime-2.1-mini costs $10.00 and $20.00. OpenAI sells GPT-Live voice sessions at $0.05 a minute, billed by the second. We read OpenAI's pricing page on October 26, 2026.
Build voice agents for trades? Get your tool listed free.
Realtime model prices (per 1M tokens)
| Model | Type | Input | Output | Cached input |
|---|---|---|---|---|
| gpt-realtime-2.1 | Audio | $32.00 | $64.00 | $0.40 |
| gpt-realtime-2.1 | Text | $4.00 | $24.00 | $0.40 |
| gpt-realtime-2.1-mini | Audio | $10.00 | $20.00 | $0.30 |
| gpt-realtime-2.1-mini | Text | $0.60 | $2.40 | $0.06 |
Source: OpenAI API pricing, checked October 26, 2026. OpenAI says the Realtime API is not priced apart from the model. Tokens bill at the chosen model's rates.
GPT-Live sessions
| Model | Price |
|---|---|
| gpt-live-1 | $0.05 a minute |
OpenAI bills GPT-Live by the second with no rounding up to a whole minute. Backend model and tool use are charged on top.
Live transcription
| Model | Price |
|---|---|
| gpt-live-transcribe | $0.017 a minute |
| gpt-realtime-whisper | $0.017 a minute |
| gpt-realtime-translate | $0.034 a minute |
Cost per minute on the token models
OpenAI does not publish a per-minute price for gpt-realtime-2.1. Your cost depends on how many audio tokens a minute of talk uses and how long the agent speaks. OpenAI's page does not state a tokens-per-minute figure. We do not guess one here. Run a test call and read the usage in your OpenAI dashboard.
The mini model's audio output costs $20.00 per 1M tokens against $64.00 for the full model. That makes the full model 3.2 times the price on output, by our math.
Phone calls need a carrier
The OpenAI price covers the AI. To answer a real phone number, you add a carrier like Twilio. Twilio charges $0.0085 a minute to receive calls on a local US number and $1.15 a month for the number. For 1,000 minutes on GPT-Live, the AI plus Twilio comes to $59.65, before backend model and tool charges. That total is our math.
OpenAI adds a 10% uplift on regional processing endpoints for models released on or after March 5, 2026.
Compare voice AI costs: Twilio voice pricing · Vapi pricing · Retell AI pricing, Bland AI pricing, Vapi vs Twilio, AI receptionist cost, Deepgram vs AssemblyAI, LiveKit pricing, SignalWire pricing.
FAQ
How much does the OpenAI Realtime API cost? gpt-realtime-2.1 costs $32.00 per 1M audio input tokens and $64.00 per 1M audio output tokens. The mini model costs $10.00 and $20.00.
Does OpenAI charge per minute for the Realtime API? Not for the token models. GPT-Live sessions are the exception at $0.05 a minute, billed by the second.
Is there a cheaper OpenAI realtime model? Yes. gpt-realtime-2.1-mini costs $10.00 per 1M audio input tokens and $20.00 per 1M audio output tokens.
How much does OpenAI live transcription cost? gpt-live-transcribe costs $0.017 a minute.
Does the OpenAI price include a phone number? No. You add a carrier such as Twilio for the number and the call minutes.
Checked 2026-10-26. Prices and policies change. Confirm on the source before you buy.
Sources
More guides
- Abby Connect pricing: plans, Abby Minutes, and AI vs human cost (2026)
- AI answering service cost in 2026: $29 to $500 a month, 8 services compared
- AI answering service for contractors: prices compared (2026)
- AI receptionist for appliance repair: 5 options from $24.95 a month (2026)
- AI receptionist for auto repair shops: 7 options from $24.95 a month (2026)
- AI receptionist for cleaning businesses: 7 options from $0 to $449 a month (2026)