O
OPENAI / gpt-live-1.1

GPT-Live-1.1

A voice model that listens while speaking. The November 19 update improves pronunciation, accent consistency and transcription in multilingual conversations. It accepts audio and text and returns both in a Live session.

01 / SUMMARY

The essential

ORIENTATIONFull-duplex multilingual voice conversation

Primary use stated or inferred cautiously from official documentation.

CONTEXTNot published

Maximum input capacity when published by the source.

EXITStreaming audio and text

Documented maximum generation limit.

ACCESSOpenAI Live API

Verified availability channels.

MODEL TYPEVoice and audio · Multimodal

Primary functional family and verified specialties.

LICENSEProprietary

Declared terms for API access, model weights, or self-hosting.

02 / TECHNICAL SHEET

Limits and integration

API IDgpt-live-1.1
Model typeVoice and audio · Multimodal
Access modelPaid proprietary
LicenseProprietary
DeploymentHosted API
Release19 NOV 2026
Knowledge cutoffNot published
EntranceAudio · Text
ExitAudio · Text
Context windowNot published
maximum outputStreaming audio and text
ReasoningNot published
Published toolsFeatures
Structured OutingsNot published
Batch processingUnavailable
Prompt cacheNot published
Fine-tuningUnavailable
Verified platformsOpenAI API
BEST SUITED FOR

Natural voice interfaces with interruptions and backend delegation for reasoning and tools.

WORTH MONITORING

Uses v1/live/sessions only: this is not the Realtime, Responses or Chat Completions endpoint. Concurrent-session limits are specific to each model and organization. No numerical context limit is published.

03 / CAPABILITIES

What it can do

01

Orientation

Natural voice interfaces with interruptions and backend delegation for reasoning and tools.

02

Context

Not published · Streaming audio and text

03

Tools and integration

Features

04

Access

OpenAI Live API

Input modalities
AudioText
04 / PRICES

Documented cost

ConceptWorthUnit/condition
Standard input0.05 USD / active minuteStandard API rate
Cached inputNot publishedReading reused prefixes
Cache write or storageNot publishedThe condition varies by provider
Standard output0.02 USD / WebRTC standby minuteMay include reasoning tokens
Batch inputNot publishedAsynchronous processing
Batch outputNot publishedAsynchronous processing

Billed by duration, not per million tokens: USD 0.05 per active minute and USD 0.02 per WebRTC standby minute with inference paused. Delegated models and tools are charged separately.

i Prices change and may depend on level, region or context length. Check the source before making a decision.

05 / EVALUATIONS

How to read the results

“

No verified quantitative score for this version.

i No verified quantitative score for this version.

Inferama only highlights a “best result” when the metric, test set, configuration, and date allow for an equivalent comparison. The supplier's figures are presented as claims from its own source.

06 / VERSIONS

Chronology

gpt-live-1.1

New voice version; standby is a session state, not another model.

07 / SOURCES

Traceability