Perplexity and Comet are both getting a serious voice upgrade today, and it quietly marks one of those moments where “talking to your computer” stops feeling like a demo and starts feeling like a habit you might actually keep.​

On the Comet side, voice mode is being reworked into something closer to a true, hands‑free copilot for the browser. You can ask Comet about whatever is on your screen, move between sites, and chat across multiple tabs without losing context, all by speaking. That means things like: “Scroll down… open the third result… summarize this PDF… compare it to the tab on the left,” without ever touching the keyboard or mouse. For anyone who lives in dozens of tabs all day, this isn’t just a convenience feature; it’s a new way to drive the browser.

Under the hood, Comet’s upgraded voice mode is now powered by OpenAI’s new gpt‑realtime‑1.5 model, which is designed specifically for low‑latency, voice‑first agents. This model improves how reliably the assistant follows instructions, how consistently it calls tools, and how naturally it responds in multiple languages. Comet says it has already optimized tool‑call stability by more than 25%, which might sound like a dry metric, but it’s exactly the kind of thing users feel immediately: fewer flaky actions, fewer “sorry, something went wrong,” and more of the assistant just doing what you asked.

Voice expressiveness is getting a noticeable bump as well. Instead of flat, robotic replies, the upgraded voices are more fluid and dynamic, with better pacing and intonation. When you’re listening for several minutes at a time—say, while it reads a long article or walks you through a workflow—that difference in tone and rhythm matters a lot more than spec sheets or token counts. Early reactions from power users have already highlighted that this kind of upgrade is the one people notice, even if no one explains what changed: tool calls stop flaking, interruptions feel smoother, and voice mode becomes something you just leave on.

Crucially, this isn’t limited to Comet’s desktop experience. The same upgraded voice capabilities are rolling out across Perplexity for all users, bringing the same real‑time feel and stability improvements to the core assistant. On mobile, that means you can talk to Perplexity instead of typing, get instant spoken answers, and keep a back‑and‑forth conversation going while you walk, commute, or cook. For people who have already been relying on Perplexity as a research or brainstorming partner, this shift toward always‑ready voice input nudges it closer to a true, everyday assistant.​

Comet’s iOS app is also part of the story. The team has confirmed that Comet for iPhone will ship with this upgraded voice mode in the coming days, after already being available on Android and desktop. iOS users can pre‑order the app now, with an expected release in March, which positions Comet as one of the more aggressive players in the “AI‑native browser” space on mobile. Combined with Perplexity’s own expansion into phone integrations and partnerships—like its recent role as an AI assistant within Samsung’s Galaxy AI ecosystem—today’s voice improvements look less like a one‑off feature and more like another step in a broader push to make conversational interfaces a default way to compute, not just an optional extra.

For everyday users, the practical implications are straightforward: fewer taps, less typing, and more talking. Students can ask Perplexity to break down a dense topic while they take notes, knowledge workers can have Comet read and compare documents while they focus on decisions, and multitaskers can move through the web with their hands busy doing something else. For developers and AI watchers, the more interesting angle is what this says about where the industry is heading: models like gpt‑realtime‑1.5 are turning voice from a novelty into an interface layer that sits on top of everything—search, browsing, apps—and glues it together with natural conversation.​

And that’s why today’s rollout matters. Taken individually, “25% better tool‑call stability” and “more expressive voices” sound like incremental updates. Taken together, across both Perplexity and Comet, they’re another sign that the future of using the web is less about pointing and clicking and more about saying what you want—and expecting your assistant to just handle the rest.