OpenAI Adds GPT-6 Astra and GPT-5.6 Controls to ChatGPT Voice

The update lets spoken conversations use more capable models for search and reasoning, while plan-based limits still determine who can use them and for how long.

By 2 min read
OpenAI Adds GPT-6 Astra and GPT-5.6 Controls to ChatGPT Voice
OpenAI Adds GPT-6 Astra and GPT-5.6 Controls to ChatGPT Voice

Listen to this story

The audio brief

About 1:21
0:001:21
Read transcript
ChatGPT Voice can now use GPT-5.6 and GPT-6 Astra for spoken requests that need web search or more demanding reasoning. The September ninth update brings Voice into the same model-control system as text chat. Instead of separate Instant, Medium, and High intelligence settings, eligible users can choose a model and adjust reasoning effort through the controls already used in text conversations. That makes the interface more consistent, but it does not make the more capable models available to everyone. Access, along with usage limits, still depends on the account plan. Astra access began with a limited group of organizations, with broader availability planned over the following days. So a spoken request can now use the same model-selection framework as a text request, but only within the boundaries of that user’s subscription. The daily limits are specific. Plus users get up to three hours of GPT-Live-1 each day. The one-hundred-dollar Pro plan allows up to fifteen hours, while the two-hundred-dollar Pro plan offers unlimited GPT-Live-1 usage. OpenAI also changed what happens at the limit: paid users are warned when they run out, rather than having Voice quietly switch the session to GPT-Live mini. The key constraint to watch is whether broader Astra access changes who can use these models in Voice—or simply makes the same plan-based controls more visible.

Story brief

3 key points

OpenAI has folded Voice into ChatGPT’s existing model-control system, letting eligible users select GPT-5.6 or GPT-6 Astra and adjust reasoning effort for search-heavy or difficult spoken requests. The September 9 change standardizes controls across voice and text, but does not broaden entitlement: model availability and limits remain plan-dependent, and Astra rollout began with limited organizations. Plus users get...

  1. 01

    Voice replaces Instant, Medium, and High intelligence settings with model and reasoning-effort controls shared with text chat.

  2. 02

    Astra access began with a limited group of organizations; broader availability was planned over the following days.

  3. 03

    Plus allowances reach three hours daily, $100 Pro reaches 15 hours, and $200 Pro offers unlimited GPT-Live-1 usage.

ChatGPT Voice can now use GPT-5.6 and GPT-6 Astra for spoken requests that need web search or heavier reasoning. OpenAI’s September 9 update gives Voice the same model and reasoning-effort controls used in text chat, but access and usage limits still vary by plan.

The change replaces Voice’s separate Instant, Medium, and High intelligence settings. Instead, people can choose a model and reasoning effort through the controls already used for text conversations. That makes the available choices easier to find across ChatGPT’s two main conversation formats, even though OpenAI still ties available models and limits to a customer’s plan.

OpenAI describes the new models as options for cases where Voice needs to search or reason through harder questions. The release does not turn Astra into a universal Voice entitlement: the company says both available models and their usage limits depend on the account plan. A person can now select the same kinds of settings in voice and text, but their plan remains the boundary around what those settings offer.

The revised daily Voice allowances

  • Plus: up to three hours with GPT-Live-1.
  • Pro at $100 per month: up to 15 hours with GPT-Live-1.
  • Pro at $200 per month: unlimited GPT-Live-1 usage.

OpenAI also changed what happens when paid users exhaust their Voice allowance. Plus and Pro accounts no longer switch to GPT-Live mini after reaching a Voice limit. ChatGPT instead alerts users when they reach the applicable limit, making the cutoff explicit rather than continuing the session through that model switch.

Astra entered this update less than a week after OpenAI introduced it for a limited group of organizations. Voice is now another place where the company can offer Astra alongside GPT-5.6 when a request calls for search or more demanding reasoning. The immediate practical change is not that every spoken exchange becomes more intensive; it is that Voice now exposes the same model-selection framework as text chat for eligible users.

Sources

  1. help.openai.comhelp.openai.com

Loading discussion...