ChatGPT Voice Mode Adds Model Selection for GPT-5.6 Sol and GPT-6 Astra
OpenAI has updated ChatGPT Voice mode to let users select specific underlying models and reasoning depths, allowing Pro subscribers to invoke GPT-5.6 Sol and GPT-6 Astra for complex tasks. Additionally, OpenAI revised daily usage limits for GPT-Live across subscription tiers and officially retired the older Instant, Medium, and High voice intelligence options. This update directly connects real-time conversational AI with OpenAI's most capable frontier reasoning models, enabling hands-free execution of deep research and complex problem-solving. It also creates clearer value differentiation across OpenAI's subscription plans, particularly for power users paying for top-tier Pro access. When complex reasoning or web search is needed, Voice mode delegates processing to the designated underlying model, counting each transfer as an individual message. Under the new rules, Go users get 3 hours of GPT-Live-1 mini, Plus users get 3 hours of GPT-Live-1, $100/month Pro users receive 15 hours, and $200/month Pro users enjoy unlimited GPT-Live-1 access without automatic downgrades.
## BACKGROUND
GPT-Live is OpenAI's real-time voice model family featuring full-duplex speech architecture, allowing users to talk and listen simultaneously for continuous, low-latency conversations. Meanwhile, GPT-5.6 Sol and GPT-6 Astra represent OpenAI's high-end foundational reasoning models designed for demanding enterprise, coding, and scientific research tasks.