On November 20, 2026, OpenAI announced the availability of GPT-Live-1.1 for developers via its API under the model ID gpt-live-1.1. The model had already been powering voice in ChatGPT Plus and Pro since October 15; development teams can now integrate it directly into their own applications, according to the official post on OpenAI's developer community forum.
What OpenAI announced with GPT-Live-1.1
GPT-Live-1.1 is the latest version of OpenAI's real-time voice model. According to the company, it delivers more natural and consistent accents across languages, greater task-completion reliability, and a new standby mode to reduce costs during extended silence. The API now includes 38 voices — seven new ones, among them calandria, hornero, and luciernaga with a Spanish-language lean, per Alpha Signal. The architecture separates the voice model, which handles real-time conversation, from the backend that manages tools and complex reasoning. Connections are established over WebRTC.
"Real-time conversational voice accessible via API marks a new chapter for customer service automation: technical and cost barriers are lowering, and what was once feasible only for large enterprises is becoming reachable for mid-sized businesses."
Davarion Group & LabsThe standby mode: what changes on cost
One reason real-time voice AI adoption has been limited is the cost of keeping sessions open while users are silent. With the new standby mode — available for both GPT-Live-1 and GPT-Live-1.1 — developers configure an idle timeout between 5 and 60 seconds. After that threshold, the session enters standby at $0.02 per minute versus the standard $0.05 per minute, a 60% reduction per OpenAI, as reported by Alpha Signal. The WebRTC call stays open and the user experience is not interrupted.
What changes for SMBs
- 01Automate phone and voice-based customer service with a model already tested by millions of ChatGPT Plus and Pro users
- 0260% standby cost reduction per OpenAI — meaningful for contact centers with low-activity periods
- 03Seven new Spanish-leaning voices — including calandria, hornero, and luciernaga — relevant for Houston businesses serving Spanish-speaking customers
- 04Concurrency limits apply per account: check your organization's limits page before sizing any deployment
- 05Integration requires WebRTC and event handling, which calls for specialized development work
For Houston businesses with phone or voice-based customer support, GPT-Live-1.1 is a concrete option for automating pre-service flows, frequently asked question responses, and intelligent call routing — without relying on costly IVR hardware. The key is evaluating session volume, silence patterns, and integration requirements specific to each operation. Outcomes vary by case.
At Davarion Group & Labs, we evaluate and implement conversational voice AI solutions for small and medium-sized businesses in Houston. If you are interested in exploring whether GPT-Live-1.1 fits your current customer service processes, we can conduct an initial assessment to identify the fit based on your specific operational context.