You may be able to try humanoid Voice Mode if you're a ChatGPT Plus subscriber

After a delay, OpenAI’s humanoid Voice Mode has finally started rolling out. The Microsoft-backed company announced on X (formerly Twitter) that the hotly-anticipated feature is now rolling out for select ChatGPT Plus subscribers.

As launched during the Spring Update event in May alongside the GPT-4o model, ChatGPT’s Voice Mode gives you more natural, real-time interactions with the ability to interrupt and respond to emotional cues within the chatbot. Like Samantha on Spike Jonze’s “Her,” sort of.

Initially announced for a June 2024 alpha release for beta testers, OpenAI said last month that it’s postponed the launch of its ChatGPT Voice Mode for a month, citing the need to enhance content detection, user experience, and infrastructure scalability.

If you want to try ChatGPT Voice Mode, you can opt-in as a ChatGPT Plus subscriber. That’ll cost about $20/month, similar to Copilot Pro’s offering from Microsoft.

OpenAI also said that over 100 external testers across multiple languages have tested the feature for safety and privacy, and a report on its performance will be available in early August. Users in this alpha phase will receive instructions via email and app notifications, with broader access planned for the fall.

As a part of the Spring Update, the Microsoft-backed company also launched the ChatGPT desktop app for Mac users a little while ago. Its Windows version, on the other hand, will come in later months, despite billions of dollars in investment that Microsoft has poured into the AI company.

Still, the app’s distribution via OpenAI’s website rather than the Mac App Store has also raised privacy concerns. OpenAI faced scrutiny over a security flaw that stored the ChatGPT desktop app’s conversations in plain text, making them vulnerable, so the app has been updated to encrypt chats despite the lack of sandboxing.