OpenAI makes the GPT-Live 1 voice model generally available in the API at $0.05 per minute

Archive entry: this covers an event from September 10, 2026, and we published it on November 8, 2026. The details come from sources dated at the time and may have changed since.

On September 10, 2026, OpenAI made GPT-Live 1, a full-duplex voice model, generally available in the OpenAI API, as announced in its launch post. It was priced at $0.05 per minute, billed per second.

What changed

• Availability: GPT-Live 1 was generally available in the OpenAI API.
• Delegation: the model supported Responses delegation or client delegation, handing work either to the Responses API or to the client.
• Pricing: $0.05 per minute, billed per second, so cost followed the length of a call.
• Highlights: OpenAI highlighted stronger instruction following, custom voices and telephony support.

What it means for people who build with AI

Per-minute pricing made voice-agent costs predictable compared with token-metered audio. For teams building voice agents, that meant a call's cost could be estimated from its length, and the delegation options let the model hand work to the Responses API or to the client.

Later updates

According to the OpenAI API changelog, GPT-Live 1 reached general availability on September 10, 2026. Prompt-to-voice custom voices followed on October 16 and outbound SIP calling on November 2. Both additions came after the launch and touched the custom voices and telephony areas that OpenAI had highlighted.

Sources

• Build more natural voice experiences with GPT-Live-1 in the API (OpenAI, September 10, 2026, official)
• OpenAI API changelog: GPT-Live 1 general availability (OpenAI, September 10, 2026, official)
• OpenAI News RSS feed (item: Build more natural voice experiences with GPT-Live-1 in the API) (OpenAI, September 10, 2026, official)

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.