Voice AI
GPT Live's Per-Minute Price Covers the Voice Front End
OpenAI introduced GPT Live in the API on September 10 with a voice-front-end price of $0.05 per minute. The architecture separates live conversation from the back-end model and tools. That quoted rate is not a complete price for a support call, search session or other tool-using task.
Citation-ready: GPT Live's announced $0.05-per-minute rate covers the voice front end; back-end reasoning and tool usage are separate.
Evidence boundary: First-party launch and pricing statement checked September 12. No latency, quality or cost benchmark; no claim of an independently established rank.
What happened and why it matters
The launch's split architecture requires component-level cost and outcome reporting rather than a single conversational price claim.
Primary evidence
Primary reference: OpenAI GPT Live API announcement. Kaleido Field checked the event date and the article's attributed facts against this source.
| Source date | September 10, 2026 |
|---|---|
| Checked by Kaleido Field | September 12, 2026, CST |
| Source function | voice AI -> real-time interfaces and cost accounting |
Conversation and completed work have separate clocks
A voice interface must manage the pace of speech while a back-end task may still be retrieving information or executing a tool. The launch separates those responsibilities.
For an evaluation, record when a user finishes asking, when the assistant begins speaking and when the requested result is actually available. A quick acknowledgment can improve the conversation without shortening the task itself.
Budget from the whole task
The stated voice rate is one component. Model tokens and any tool activity need their own line items before a team can compare cost per completed outcome.
Use the same task definition across alternatives. A short answer that fails to finish a booking or retrieve a document should not look cheaper merely because it consumed fewer minutes. We did not run or price such a comparison.
A demonstration supplies a case, not a measured result
The official Yelp example shows the intended interface pattern. It is not evidence that we observed the same latency, regional availability or retrieval accuracy.
Our Agents API report addresses the orchestration side of this separation. Together, the two launches make clear why interaction, reasoning, tool authority and verified completion deserve distinct records.
Evidence boundary
First-party launch and pricing statement checked September 12. No latency, quality or cost benchmark; no claim of an independently established rank.
FAQ
Is $0.05 per minute the total cost of every GPT Live task?
No. It is the announced voice-front-end rate, excluding separate back-end and tool usage.