- Published on
OpenAI ChatGPT Voice Mode Supports Calling GPT-5.6 Sol and GPT-6 Astra
- Authors
- Name
- DeepHorus
On September 10th, OpenAI announced a major update to its ChatGPT Voice mode, officially introducing a model selection feature. This upgrade allows users to freely choose the desired text model and reasoning depth during voice interactions, with Pro plan users having the additional option to invoke advanced GPT-5.6 Sol or GPT-6 Astra models. This move marks a significant step in the evolution of voice assistants from single-model to more flexible and powerful multi-model architectures.
According to Atty Eleti, OpenAI's head of voice products, when performing web searches or complex reasoning tasks, the voice mode will automatically invoke the model pre-specified by the user. The core of this mechanism involves the voice mode converting the user's query into a text message, which is then sent to the selected underlying text model for processing, seamlessly integrating voice input with the capabilities of powerful text models.
Alongside this feature upgrade, OpenAI has also simplified and clarified the daily usage quota rules for GPT-Live. Under the new scheme, Go plan users receive up to 3 hours of GPT-Live-1 mini usage, but no longer have access to GPT-Live-1. Plus plan users, on the other hand, are allocated up to 3 hours of GPT-Live-1 usage.
The rules are even more favorable for higher-paying Pro users. Pro users paying $100 per month receive up to 15 hours of GPT-Live-1 usage, while top-tier Pro users paying $200 per month enjoy unlimited GPT-Live-1 access. This reflects OpenAI's differentiated service strategy catering to different user tiers.
Notably, after this update, once Plus and Pro users exhaust their respective voice interaction quotas, they will no longer be automatically downgraded to the less capable GPT-Live mini model. This ensures that paid users consistently experience the advanced models matching their plan level until their quota is fully utilized.
Simultaneously, OpenAI announced that the original three-tier independent voice intelligence settingsā"Instant / Medium / High"āhave been officially discontinued. They are replaced by the new method of users directly selecting the underlying text model, making the control of the voice assistant's intelligence level more intuitive and specific.
This series of updates is expected to further enhance the practicality and appeal of ChatGPT Voice mode. By granting users greater model selection authority, OpenAI aims to elevate the flexibility and performance of voice interactions to new heights, potentially solidifying its leading position in the AI voice assistant domain.
