Use this GPT Live pricing calculator to estimate a documented gpt-live-1 voice session from exact seconds. The calculator adds optional backend model and tool charges as separate line items, keeps transcription separate, and uses the current ChatGPT Voice pricing and limit sources for context.
The Live API duration tool is the active calculation. Realtime token rates are documented separately below; the current API calculation uses gpt-live-1 session seconds, not Realtime token rates.
Enter a planning usage pattern only. OpenAI's current ChatGPT pricing and Voice FAQ provide the consumer price and limit context; this panel estimates entered voice time and makes no plan recommendation.
Consumer access boundary: This panel estimates entered voice time only. It does not infer a model assignment or recommend a plan; check the current official pricing and Voice FAQ for the account you intend to use.
This control calculates the current gpt-live-1 Live duration charge from exact seconds. It does not calculate Realtime token usage; the current Realtime audio and text rates are documented in the reference section below.
Live duration rate, rechecked September 20, 2026: gpt-live-1 voice sessions cost $0.05 per minute billed per second; backend model and tool usage is separate. gpt-live-transcribe is $0.017 per minute and is not mixed into the Live voice total. Realtime token rates are listed in the reference section below. Sources: gpt-live-1, gpt-live-transcribe, and API pricing.
The page separates a consumer usage planning view from a developer session-cost view. The API calculation uses exact seconds × $0.05 / 60 for gpt-live-1, then adds optional backend model and tool charges. It does not project a monthly bill, and consumer subscriptions do not establish API credit.
Minute-based estimates can hide exact-second billing. The current GPT-Live-1 calculator uses exact session seconds and does not invent an audio-token-per-minute conversion. Enter the session duration that you actually expect to bill.
The current result shows voice duration, backend model charges, and tool charges as separate lines. It reports the entered session cost only; it does not silently multiply the amount into a monthly or other period projection.
Token-mode estimates do not describe GPT-Live-1 voice duration. Calculate voice cost from seconds, then enter backend model and tool charges as independent USD values so the displayed total remains auditable.
The result covers the entered live voice seconds plus the optional backend and tool values. It does not estimate transcription, telephony, infrastructure, storage, network, taxes, or other charges. Use the OpenAI billing dashboard and invoice as the final record.
The verified gpt-live-1 voice rate is $0.05 per minute, billed per second without rounding up. Backend model and tool usage is charged separately. The verified gpt-live-transcribe rate is $0.017 per minute and belongs in a separate transcription estimate. Recheck the official Developer pricing source before committing a production budget.
| Model | Billing unit | Verified rate | Separate charges |
|---|---|---|---|
| gpt-live-1 | Live voice duration | $0.05 / minute, billed per second | Backend model and tools |
| gpt-live-transcribe | Transcription duration | $0.017 / minute | Keep separate from live voice |
Realtime token reference: the current official GPT-Realtime-2.1 model entry lists audio at $32.00 input / $0.40 cached input / $64.00 output per 1M tokens and text at $4.00 / $0.40 / $24.00 per 1M tokens. It identifies gpt-realtime-2.1 as the model choice and lists the v1/realtime endpoint. The Free API tier is unsupported for this model. These are Realtime token charges; they do not replace the gpt-live-1 Live duration rate used by the calculator.
The current official model entry uses the name gpt-realtime-mini. It lists text at $0.60 input / $0.06 cached input / $2.40 output per 1M tokens, audio input and output modalities, and v1/realtime. The checked entry does not publish a separate audio-token price, so this page leaves that field unfilled instead of carrying forward an older mini rate. Use the official model page when choosing between the current Realtime entries.
The Consumer Plans tab estimates entered voice time by multiplying conversations by average minutes. It does not claim current published plan limits, prices, model assignment, or eligibility because those details require current official product information. Check the current official consumer information for the account you intend to use.
Current consumer plan access, price, model assignment, and limits require current official product information. The cards ask you to check that information instead of presenting historical values as guarantees.
Current Pro plan pricing, limits, and guardrails require current official product information. No lowest-cost consumer recommendation is generated by this calculator.
These examples show the exact-second arithmetic used by the GPT-Live-1 session calculator. They keep sub-cent values visible and keep backend or tool charges separate from voice duration.
For a verified live session, 60 seconds × $0.05 / 60 = $0.05. A one-second session is approximately $0.000833, so it is not free merely because a display rounds currency to cents.
Do not convert GPT-Live-1 duration into an invented audio-token-per-minute rate. Add any backend model or tool charges as separate line items.
Token estimates are not used by the current GPT-Live-1 calculator. For a 90-second live session with $0.02 backend work and $0.01 tools, voice is $0.075, the separate charges are $0.02 and $0.01, and the total is $0.105.
Use the billing dashboard and invoice for final reconciliation. Transcription has its own documented rate and should be calculated separately.
Consumer subscriptions and developer API usage are separate products. The verified gpt-live-1 API calculation is session seconds × $0.05 / 60, plus separately reported backend model and tool charges. A consumer plan does not become API credit, and this page does not publish a consumer-versus-API break-even claim.
Use current official consumer information for plan decisions and the API billing dashboard for production estimates. Keep transcription, telephony, infrastructure, taxes, and other costs outside the live voice line item unless you explicitly add them.
The current gpt-live-1 Live voice rate is $0.05 per minute billed per second, with backend model and tool usage charged separately. Realtime models use their own official token rates; the Live duration rate is not a Realtime token rate.
Current consumer plan price, model assignment, and limits require official product information for the account and surface you use.
Multiply non-negative finite gpt-live-1 session seconds by $0.05 / 60. Then add backend model and tool charges as separate fields. Do not round seconds up or mix the $0.017 per-minute transcription rate or the separate Realtime token rates into the Live duration total.
A consumer plan value judgment requires current official price, entitlement, and limit evidence. This calculator does not make that recommendation; API charges remain separate.