OpenAI has expanded its GPT Live voice mode to accept file attachments and tie into projects, letting users bring documents and other material directly into spoken conversations. The update, which rolls out to users now, extends the voice assistant's reach beyond simple back-and-forth chat.
What the update changes
Voice mode on GPT now works with files you attach during a session. Instead of just talking, you can hand the assistant a PDF, an image, or another document, and it can refer to that content while responding. The feature also connects to projects, meaning the assistant can pull context from work you've already organized under a project umbrella.
That's a meaningful shift. Previously, voice mode was largely a hands-free way to ask questions or get quick answers. With file support, it becomes a tool for working through real material—contracts, notes, screenshots—without needing to switch to a text interface.
Why the integration matters
The project tie-in is the bigger piece. By linking voice conversations to projects, OpenAI is positioning the assistant as something that tracks ongoing work, not just one-off interactions. You could, for example, discuss a draft document during a voice call and have the assistant remember that context in later sessions.
That kind of continuity is what separates a useful voice assistant from a novelty. It also suggests OpenAI is pushing toward a more unified workspace, where voice, files, and projects all live in the same loop.
How it fits with GPT's broader design
OpenAI has been steadily adding capabilities to GPT, and this move fits that pattern. The company has already built multimodal features into its models, and voice mode has been a key part of the consumer app. Adding file attachments closes a gap between what you can do by typing and what you can do by speaking.
It's not hard to see why they'd prioritize this. Voice is convenient, but it's limited when you can't reference something on screen. Now that limitation is gone.
What's still unknown
OpenAI hasn't said whether the expanded voice mode will be limited to certain subscription tiers or whether it works across all platforms. The rollout appears to be live, but the company hasn't detailed any phased availability or regional restrictions.
The update is available now, and users can test it by starting a voice session and attaching a file. How far OpenAI takes this—whether it adds more file types, deeper project controls, or collaborative voice sessions—remains an open question.
