GPT-Live-1.1 Enters the API: Validate Accents, Standby Billing, and Delegation
Use one fixed conversation to produce separate receipts for voice, WebRTC standby, and delegated work
OpenAI added GPT-Live-1.1 to the Live API on November 19, 2026, with full-duplex multilingual voice and a standby state for WebRTC sessions. This guide does not treat release notes as a hands-on result. It provides one acceptance card: run a fixed script for accent handling, numbers, interruption, noise, and recovery; measure active and standby minutes separately; account for delegated backend work as a separate cost; and verify that the application records permissions, confirmations, task state, and the result returned to the live conversation. The goal is to decide whether the model fits a real voice-agent workflow, not merely whether a connection can be opened.
Key takeaways
By ENHE AI · 2026-11-20 · AI-assisted drafting; pre-publication quality checks completed.
Direct answer: connection success is not acceptance
GPT-Live-1.1 is in the Live API, but full-duplex and multilingual labels do not replace testing with your accents, devices, and noise. Use the same non-sensitive conversation for every candidate configuration.
Record first-audio latency, omissions, numbers, names, interruption, noise, and reconnect behavior. Attach time, client version, and recovery action to each failure instead of writing only “poor experience.”
Validate standby billing separately
The model page lists different per-minute prices for active inference and WebRTC standby. Standby means paused inference in a WebRTC session, not every idle client moment or a backend task that is still running.
Log session start, standby entry, standby exit, and close time, then reconcile them with usage records. If the phases cannot be separated, do not put the estimate into a customer quote.
Delegation must return a verifiable result
A live conversation can delegate through the Responses API or a client-managed path, but the application still owns permission, confirmation, task state, and records. The agent needs an explicit downstream success result before saying work is complete.
Model timeout, cancellation, retry, and partial completion as separate states. Keep human confirmation for payment, account, production-data, or external-message actions; fluent speech must not hide incomplete work.
Completion criteria and fallback
Completion means the fixed script has itemized results, billing phases reconcile, delegated actions have state and receipts, and an older model or text route remains available. Passing only one part is partial.
Start with a small low-risk internal sample before expanding. This article did not call the API, generate audio, or validate an account bill.
FAQ
Is standby simply a connected client doing nothing?
Not necessarily. The documentation limits standby to paused inference in a WebRTC session, so verify session state and usage records.
Can a voice agent announce a delegated task as complete immediately?
No. The application should first receive a verifiable downstream result and retain permission, confirmation, and task state.
Fact evidence
- GPT-Live-1.1 is a full-duplex multilingual voice model for the Live API. — OpenAI Developers — GPT-Live-1.1 model
- Standby applies only to paused inference in a WebRTC session. — OpenAI Developers — GPT-Live-1.1 model
- Backend work delegated from a live session is billed separately. — OpenAI Developers — GPT-Live-1.1 model
- The application owns permissions, confirmations, records, and task state when a live agent delegates work. — OpenAI Developers — Live API delegation
Related tools and reading
These production pages support the same decision, workflow, or safety boundary.
Conclusion
GPT-Live-1.1 is valuable only when voice behavior, standby cost, and delegated actions form one auditable chain. Collect evidence with a fixed acceptance card before scaling.
What this means for everyday users
Completion means the fixed script has itemized results, billing phases reconcile, delegated actions have state and receipts, and an older model or text route remains available. Passing only one part is partial.
Related reading
A voice-agent release checklist for VAD, noise, interruptions, transcription, tools, and failure handoffs
A voice-agent release should not be accepted after one successful conversation in a quiet room. Test cases need to cover speech-activity detection start and stop behavior, silence and background noise, user interruption, transcription evidence and correction, spoken status around tool calls, and state recovery after timeouts or failures. The OpenAI Realtime API reference documents capabilities for real-time sessions, audio input and output, transcription, and server-side voice activity detection. This article turns those capabilities into a general acceptance checklist. It is not a release announcement, a claim about a particular product version, or a performance guarantee. Teams should define their own language, device, privacy, latency, escalation, and human-review criteria before deployment.
Eleven Realtime v1 Mini and Max Launch: Choosing a Voice-Agent Path
ElevenLabs launched Eleven Realtime v1 Mini and v1 Max on October 20, 2026. Both speech-to-speech models support more than 90 languages and over 17,000 voices, but they target different tasks: Mini emphasizes lower-cost natural conversations, while Max is designed for support, lead qualification, and multi-step work. This article offers a selection and migration checklist instead of repeating benchmark claims. It explains how to compare both models on the same script, what OpenAI Realtime API compatibility means, why WebRTC integrations must move to WebSocket, and when ElevenAgents is the better starting point for channels, knowledge bases, guardrails, testing, and monitoring.
ChatGPT Will Retire Advanced and Standard Voice: Run a Migration Check Before December 11
OpenAI says Advanced Voice and Standard Voice will retire on December 11, 2026. Ordinary voice conversations will then default to GPT-Live, and Advanced Voice inside custom GPTs is deprecated on the same date. This guide provides a migration rehearsal rather than assuming the default switch is equivalent. Inventory devices, languages, headsets, custom GPTs, video, and screen sharing; then run one fixed script covering startup, interruption, numbers, names, background noise, and a longer exchange. Record feature gaps and rollback routes for each surface. Live video and screen sharing remain on Advanced Voice until GPT-Live supports them, so a visible voice button alone does not prove migration is complete.
From Chat Boxes to Personal AI Companions: AI Assistants Are Entering the Desktop Execution Era
AI assistants are moving from answering questions toward continuing real tasks. AI agents, MCP tool ecosystems, personal memory, and local workbenches are pushing this shift together. For users, the real value is not another chat box, but less repeated context setup and more continuity from thinking to doing.
ChatGPT Adds an Accounting Plugin: Keep Source, Variance, and Approval Tables for Month-End Work
OpenAI announced the ChatGPT Accounting plugin on November 18, 2026. It can help reconcile transactions, draft invoice entries, and prepare month-end reports from records and templates, while Enterprise workspaces still require administrator approval before installation. This article does not provide personalized accounting or financial advice. Instead, it supplies three control tables: a source table for files, periods, and versions; a variance table for every unmatched amount; and an approval table that names who may confirm, export, or roll back a change. Bank access, payments, credentials, and final posting remain human-reviewed steps. The plugin output is a reviewable draft, not a closed accounting period or evidence that the underlying records are complete.
GitHub Copilot Evals Enters Preview: Regression-Test Skills, Agents, and MCP Changes
GitHub launched Copilot Evals in public preview on November 17, 2026. It creates repeatable tests for skills, custom agents, and MCP tools and lets users compare runs before and after a customization change. This guide defines a minimum regression set: choose three realistic tasks, then specify the input, allowed tools, success criteria, and prohibited behavior for each. Preserve at least one baseline run and one post-change run, including the result, execution path, failed checks, and human review. Merge a customization only when the change preserves the original task, permission boundary, and evidence chain. A single passing answer is not enough because model output can vary, and an automated check cannot decide every domain-specific quality or safety question.
