AI Launch Profile
GPT-Live launches in ChatGPT Voice; API access remains pending
OpenAI began rolling GPT-Live-1 and GPT-Live-1 mini into ChatGPT Voice for users globally. The voice model uses full-duplex audio for simultaneous listening and speaking, and it can delegate search, reasoning or longer work to a separate frontier model while keeping the conversation active. The July 8 announcement says API access is planned soon; it is not part of this launch.
At a glance
Launch Snapshot
- Company
- OpenAI
- Launch date
- July 8, 2026
- Launch type
- Model Release
- Category
- AI Models, AI Voice/Audio Tools
- Audience
- Consumers, Creators, Developers, Product Teams
- Pricing
- OpenAI did not announce a standalone GPT-Live price. Its July 8 release assigns GPT-Live-1 to ChatGPT Go, Plus and Pro and GPT-Live-1 mini to Free. Current ChatGPT Voice documentation lists separate desktop allowances by plan and states that Desktop Voice is not available via API key; verify the live plan page because limits and surfaces can change.
- Free plan
- Yes
- API
- No
- Open weights/source
- No
Launch Context
Use these links to move from this record into the broader Launch Intelligence database.
Verification & Sources
- Status
- Verified
- Source links
- 4
- Freshness
- Verified July 29, 2026
- Last verified
- July 29, 2026
- Last updated
- July 29, 2026
Key source checks
Suggest a correction
Creator Coverage Next Steps
This launch has signals that may support demos, reviews, creator education, founder storytelling, or practical product explainers.
Launching an AI product that needs clear demos, creator education, and buyer trust? Sponsor a Kingy AI video or launch feature.
Kingy AI Take
GPT-Live’s important change is architectural and experiential: continuous interaction is separated from deeper task execution. That could make voice useful for steering work rather than merely dictating prompts. The evidence supports a ChatGPT launch, not a new API product. Teams should assess interruption handling, accents, noisy environments, delegated-task visibility, privacy and plan limits before adopting voice for consequential work.
Who it is for
ChatGPT users who want lower-friction spoken interaction, interruptions, pauses and delegated work. Developers evaluating API use should remain on OpenAI’s existing documented voice-agent surfaces until GPT-Live API availability is officially announced.
What feels promising
Full-duplex turn-taking and background delegation address two recurring voice-assistant failures: rigid wait-for-silence turns and conversations that stall while a harder task runs.
What feels unproven
Kingy did not run a controlled GPT-Live comparison. OpenAI’s preference, reasoning and telecom evaluations are first-party; real performance can vary by language, environment, account, rollout, workspace setting and delegated task.
Source-backed record