GPT-Live-1 enters the voice API
OpenAI announced API access to GPT-Live-1 for full-duplex voice experiences.
Announced: 2026-09-10 · Published on Xavkit: 2026-10-16 · Last editorial update: 2026-10-16
What happened
OpenAI announced API access to GPT-Live-1 for full-duplex voice experiences. Its focus includes conversational interruption handling and controls for spoken interactions. Real deployments still require robust audio, consent, privacy, and tool fallback designs.
Why it matters
Live voice quality depends on the whole conversation loop, not just the speech model. Accent variation, background noise, interrupted turns, network behavior, and accessible alternatives should be evaluated.
Official sources
What changed
• GPT-Live-1 was introduced in the API
• The release emphasizes full-duplex speech
• Developer voice controls were expanded
Who it affects
• AI product users
• Developers and technology teams
What to do now
• Test noisy environments and interruptions
• Evaluate consent and fallback
• Measure end-to-end latency
Related updates
GPT-6 Sol and Luna broaden model choices
OpenAI introduced GPT-6 Sol and Luna as alternatives with different capability and efficiency balances.
GPT-6 and Intelligent UI in ChatGPT
OpenAI introduced interactive response elements in ChatGPT with the GPT-6 rollout.
GPT-6.1 Sol and professional task workflows
OpenAI announced GPT-6.1 Sol as a new model option for coding and professional workflows.
ChatGPT Voice follows instructions better (and repeats less).
Voice is less goofy and more obedient now.
Claude API: structured outputs support expands, and older models keep getting retired.
Production users: pin versions and stop ignoring deprecation notes.