SIAgentNews.com
← All news
DevelopingOpenAI

OpenAI to retire legacy ChatGPT voice modes by December 2026

OpenAI is phasing out Advanced and Standard voice modes in favor of GPT-Live, which allows for more natural, simultaneous conversation.

· First reported by Interesting Engineering on MSN

nearly doubledvoice usage since July

The short version

  • OpenAI is retiring Advanced and Standard voice modes on December 11, 2026.
  • Voice usage on the platform nearly doubled after GPT-Live was introduced in July.
  • GPT-Live allows for simultaneous listening and speaking, removing the need for turn-taking.
  • Custom GPTs using older voice technology will also lose access to those models in December.
  • Live video and screen-sharing will continue to use the Advanced voice infrastructure for now.

Why it matters

This shift marks a move toward more human-like interaction where AI can handle complex tasks in the background without breaking the flow of conversation. Users who rely on custom GPTs or specific voice workflows will need to adjust to a unified system by late 2026.

The shift to GPT-Live

OpenAI has announced plans to discontinue two of its legacy voice options, Advanced and Standard, from the ChatGPT platform. These modes will officially disappear from the model picker on December 11, 2026. The company is moving toward a unified experience where GPT-Live serves as the primary voice interaction method.

The transition comes as a result of significant user adoption of the newer technology. According to reports from Interesting Engineering on MSN, nearly all ChatGPT voice users have already switched to GPT-Live since its debut in July. During that same period, overall voice usage on the platform nearly doubled.

Users who currently prefer the older conversational modes will have time to adjust before the deadline. OpenAI has begun notifying users about the retirement as of November 10, but the existing settings will remain functional until the end of the transition period in 2026.

How the new architecture works

GPT-Live represents a fundamental shift in how the AI processes audio compared to its predecessors. Unlike the older systems, GPT-Live can listen and speak at the same time. This architecture removes the requirement for conventional turn-taking, where a user must finish speaking before the AI begins its response.

This capability allows for a more fluid experience, including the ability to handle natural interruptions from the user. The model can also provide brief acknowledgments while a user is still talking or remain silent if it detects the user needs time to think. These features are designed to make the interaction feel less rigid and more like a human conversation.

Furthermore, GPT-Live is designed to connect with OpenAI's text-based models to handle more demanding requests. It can delegate complicated, multi-step tasks to these models in the background while maintaining the ongoing voice conversation. This means the system can perform heavy lifting without ending the active audio stream.

Comparing the legacy systems

The older modes functioned using different technical approaches that OpenAI now intends to consolidate. Advanced voice processed incoming and outgoing audio directly, but it was limited by a sequential flow. It had to wait for a user to complete their thought before it could generate any response.

Standard voice was even more complex, relying on three separate models to function. One model converted the user's speech into text, a second generated a written response, and a third converted that text back into audio. This multi-stage process often resulted in noticeable delays and the potential loss of information during the conversions.

By retiring these options, OpenAI aims to streamline the experience into a single, high-performance voice system. The company plans to continue improving the depth of responses and conversational continuity as GPT-Live develops. This consolidation is intended to provide a more cohesive experience for all users regardless of their specific needs.

Impact on custom GPTs and features

The retirement of Advanced and Standard voice also has specific implications for developers and power users. Custom GPTs that currently utilize the older voice technology will lose access to those models on the December 11 deadline. Developers will need to account for this change when planning their existing voice-based workflows.

However, there is a notable exception for interactive visual features. Live video and screen-sharing will continue to run through the Advanced voice infrastructure for the time being. GPT-Live does not currently support these specific capabilities, and OpenAI has not provided a timeline for when they will be integrated.

This creates a temporary distinction between standard voice conversations and visual interactions within the app. While most users will see a move toward GPT-Live, those who rely on video or screen-sharing will see their current infrastructure remain intact. OpenAI will maintain the older system until the newer model gains equivalent capabilities.

What to expect next

Users do not need to take any manual action to migrate their data or settings. After the December 2026 deadline, ChatGPT will automatically switch affected users to GPT-Live. Previous conversations will remain accessible in the history, and users can continue those discussions through text or the new voice experience.

The transition period provides a long window for users to become accustomed to the new interaction style. During this time, OpenAI will focus on refining the personalization and depth of the GPT-Live system. The goal is to ensure that the move to a single voice experience does not compromise the quality of the interaction.

As the deadline approaches, the distinction between the old and new models will fade as the legacy options are removed from the settings menu. This move signals a clear direction for the platform, prioritizing a more dynamic, simultaneous communication style over the older, sequential models.

What to watch

  • When GPT-Live will receive support for live video and screen-sharing.
  • How developers will adapt custom GPTs to the new unified voice architecture.

Try it

Check your ChatGPT settings under the Voice section to see which model you are currently using before the December 2026 deadline.

Sources and links

First reported by Interesting Engineering on MSN. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

More AI news