On 8 July 2026, OpenAI announced GPT-Live-1 and GPT-Live-1 mini. GPT-Live continuously processed input while generating speech and could keep the conversation moving while delegating deeper work to GPT-5.5.
This release brief was checked against first-party material on 10 August 2026. The date above is the public announcement date, not the date a repository was created or a third-party provider added the model. Where access or weights arrived later, that distinction is recorded below.
Release record
| Field | Verified detail |
|---|---|
| Announcement | 8 July 2026 |
| Availability or weight release | 8 July 2026 in ChatGPT Voice, with API access announced as forthcoming |
| Release type | proprietary full-duplex realtime voice-model family |
| Access | ChatGPT product rollout at launch; API notification programme; no open weights |
| Architecture | full-duplex continuous audio interaction with delegated frontier reasoning and tool work |
| Maximum stated context | not stated as a standalone token limit in the launch article |
What changed
The model replaced turn detection based only on silence with many interaction decisions per second: speak, listen, pause, interrupt or call a tool. OpenAI launched two named checkpoints and documented a separate safety system for live speech. That is a material model and architecture boundary, not merely a ChatGPT interface refresh.
The practical comparison is therefore not simply whether GPT-Live-1 and GPT-Live-1 mini has the largest headline score. Teams need to ask whether its architecture, access terms, latency, tool behaviour and evaluation setup match the workload they actually intend to run. A model can lead one harness while losing on cost, refusal behaviour, multilingual quality or repeatability in another.
Benchmarks worth retaining
| Evaluation | Reported result | How to read it |
|---|---|---|
| Human conversation preference | GPT-Live-1 and mini strongly preferred to Advanced Voice Mode | Matched 5–10 minute conversations covering flow, interruptions and naturalness |
| BrowseComp | strong gain over Advanced Voice Mode | Agentic web-search evaluation reported at launch |
| Tau3-Voice Telecom | outperformed Advanced Voice Mode | Internal variant using a customised simulated user model |
These are release-time results, not independently reproduced guarantees. OpenAI reported preference and task gains without publishing every chart value in text, and the delegated frontier model can change over time, so the serving stack must be recorded with each evaluation. Scores should remain attached to the disclosed effort setting, agent harness, tool access, timeout, context-management policy and judge model. Moving a number into a procurement sheet without those conditions creates false comparability.
Architecture and access
GPT-Live-1 and GPT-Live-1 mini is described as full-duplex continuous audio interaction with delegated frontier reasoning and tool work with not stated as a standalone token limit in the launch article of stated context. Its access position at verification time is ChatGPT product rollout at launch; API notification programme; no open weights. That wording matters: open weights, source-available weights, an API, a product preview and a research demonstration give adopters very different rights and different levels of reproducibility.
Before deployment, record the exact model identifier or checkpoint, inference stack, quantisation, reasoning setting, region, price schedule and supplier terms. If the release uses a custom licence, read the licence itself rather than relying on the word “open” in launch copy. If it is API-only, preserve the dated documentation and change-notice route because the served snapshot can change without a downloadable artefact.
What an evaluation should test next
For GPT-Live-1 and GPT-Live-1 mini, a credible internal gate should include:
- a frozen set of representative tasks with pass, fail and abstain criteria;
- a matched baseline using the same tools, timeout, prompt budget and reviewer rubric;
- repeated runs to expose variance rather than reporting a single best attempt;
- latency, token use and total task cost alongside task success;
- adversarial, multilingual and long-context cases relevant to the real deployment; and
- rollback evidence showing the previous model can be restored safely.
The wider model change-control guide explains how to keep model, prompt, tool and corpus changes reconstructable. The AI dependency inventory guide covers the release and supplier records needed after deployment.
AIEngine verdict
GPT-Live is a major omission from the earlier register. Voice teams should test interruption timing, background noise, delegated-task completion, safety steering and watermark verification as one end-to-end system.
This is a launch assessment, not a certification. Benchmark leadership is useful evidence of where to test; it is not authorization to place the model in a high-impact workflow without domain evaluation, security review and an accountable owner.
Primary sources
Image provenance
Hero image: OpenAI official release artwork. The locally served WebP is a crop of the first-party release or model-card asset recorded in the repository provenance manifest.
