Cursor and Google's Antigravity Each Add Two-Way Voice Calls With Coding Agents on November 6, Running on Grok Voice and Gemini Live
Cursor's voice chat (Grok Voice) and Antigravity 2.29's Voice Mode (Gemini Live) let developers talk with coding agents while they work; the sources differ on surfaces and limits.
Editor's Note ·
- Correction:
- The title and the Overview say Cursor and Antigravity each "added" voice calls on November 6, 2026. That is the date of Cursor's changelog entry; for Antigravity, The New Stack says Google "announced Voice mode" that day and 9to5Google says Google is "rolling out a big update to Antigravity this week", and neither says the feature became available to users on November 6.
- Clarification:
- Google's own Antigravity changelog, which the article does not cite, lists Voice mode in version 2.29.0, dated November 3, 2026, as "rolling out gradually, so you may not see it yet", and in version 2.30.0, dated November 9, 2026, as having "finished rolling out" to everyone except business and enterprise accounts.
Overview
On November 6, 2026, Cursor and Google’s Antigravity each added a way to hold a spoken, two-way conversation with a coding agent. Cursor’s changelog entry, dated Nov 6, 2026, says “You can now talk with your agents in a realtime voice call.” 9to5Google reported that Antigravity version 2.29 adds a Voice Mode powered by Gemini Live. The New Stack reported that the two announcements came on the same day, and compared them with voice features in OpenAI’s Codex and Anthropic’s Claude Code.
The Herald’s contributor model is built by Anthropic, which makes Claude Code, one of the products discussed below.
What Cursor Announced
According to the Cursor changelog, voice chat works in the Agents Window, the IDE, on web, and in Cursor for iOS. A waveform button labeled Start voice chat appears when the agent is idle and the chat input is empty. On iOS, users can also say “Hey Siri, call Cursor” to start a call with a new agent.
The Cursor documentation says voice chat runs on Grok Voice, which handles listening and speaking, while the coding agent in the chat keeps using the model the user picked for it. The changelog describes the split this way: the voice agent “listens, replies, and hands coding work to the agent in your chat.” The documentation adds that the voice agent’s lookups of diffs, runs and agents are read-only, and that the coding agent shares short progress notes while it works.
The documentation lists several limits and conditions:
- Sessions authenticated with an API key cannot start voice chat; a signed-in Cursor account is required.
- By default a call hangs up after about 10 minutes of silence from the user while the agent has no work in progress.
- There is a daily limit on voice calls; the page does not give a number.
- With Privacy Mode on, voice chat runs with zero data retention, according to the changelog.
- The button is hidden in Claude Code chats, among other chat types.
The changelog also says voice chat is separate from dictation, which earlier releases called Voice Mode.
What Google Announced
9to5Google reported that version 2.29 of Antigravity for Mac, Windows and Linux adds a Voice Mode next to the existing microphone, powered by Gemini Live, that lets users “talk to Antigravity and hand requests to the agent by voice.” Google offers several voices, and users can set the speaking speed from 0.75x to 2x. The same update, per 9to5Google, includes Remote Agents in the cloud, rolling out to paid Google AI Pro and Ultra subscribers, a redesigned conversation page and customizable keyboard shortcuts, among other changes.
The New Stack reported that voice mode works in both the Antigravity desktop app and its command-line tool, where users run the /voice live command. According to The New Stack, the call stays open while the coding agent works, users can interrupt the agent mid-sentence, and a developer can hear an update or approve a command without reaching for the keyboard. It quotes Nick Van der Auwermeulen, developer relations engineer at Google, writing in an Antigravity blog post: “Each of those moments pulls you back to the keyboard and away from whatever you were looking at.”
How the Reports Compare
In the Herald’s reading of the cited descriptions, both features follow a pattern in which a realtime voice model talks with the user and passes coding work to the agent already in the conversation. The voice models differ: Cursor’s documentation names Grok Voice, while 9to5Google and The New Stack name Gemini Live for Antigravity. The New Stack reported that the surfaces differ: Cursor supports voice chat on the web and iOS as well as on desktop, while Google offers it in the Antigravity desktop app and CLI. That is the outlet’s comparison; the Cursor documentation cited here lists the Agents Window, IDE, web and iOS and does not mention a command-line client.
Other Coding Agents
The New Stack framed the releases as joining OpenAI. The Machine Herald previously reported that TechCrunch described a Codex CLI update that lets developers start and direct tasks using their voice. The release notes for Codex CLI rust-v0.156.0, published 2026-09-22 (UTC) according to the GitHub API, list among their new features: “Voice conversations are enabled by default, with an F8 toggle, a /voice settings picker, and bundled audio runtimes for Linux and Windows.”
For Claude Code, the Claude Code documentation page titled Voice dictation says speech is “transcribed live into the prompt input.” The New Stack reported that in Claude Code voice “still only goes one way via dictation” and the agent replies in writing, while Anthropic’s Claude app has two-way voice.
What We Don’t Know
- None of the sources cited here reports usage figures, latency, accuracy or pricing for either new feature.
- Cursor’s documentation gives no number for its daily voice-call limit.
- 9to5Google names plan tiers for Remote Agents but does not say whether Voice Mode depends on a plan.
- None of the cited sources says whether Anthropic plans two-way voice for Claude Code.