Google Antigravity adds Gemini 4 Carbon and previews /swarm multi-agent orchestration

Available in preview on Google AI Pro and Google AI Ultra across the Antigravity desktop App and CLI, the /swarm command coordinates parallel coding workers, independent diff verifiers and a stall monitor over one shared zero-idle-cost message board.

Google Antigravity added Gemini 4 Carbon for Google AI Ultra subscribers across its desktop App, CLI, and IDE extensions on 10 November 2026, while launching /swarm in preview for Google AI Pro and Google AI Ultra plans in the App and CLI. Arriving five days after Antigravity’s Remote Agents rollout, running /swarm turns the active session into a coordinator that breaks complex tasks across named parallel workers.

Shared board, isolated workspaces and mixed-model roles

Workers edit concurrently inside shared git-worktree-style working copies or isolated repository branches while communicating through one append-only message board using targeted @name, @group, and @all mentions. Waiting agents sleep on the board at zero token cost until mentioned. Each swarm pairs builders with a read-only verifier agent (typically one verifier per 2 to 4 workers) that runs test suites and audits diffs before work merges, plus a lightweight monitor agent that nudges stalled steps.

Antigravity Swarm Graph tab visualising message dependencies, active workers and verifier checkpoints across a multi-agent coding run — The Swarm side pane’s Graph tab replays task handoffs, worker branches and verifier sign-offs in real time.Image: Google

Developers inspect live runs through the App’s Swarm side pane (Board, Members, and an interactive Graph timeline) or via /swarm board in the terminal, and can pin Gemini 4 Carbon to coordination and verification alongside lighter worker models after five weeks of internal Google dogfooding.

Technical details

Eligible subscriptions

Google AI Pro and Google AI Ultra for /swarm preview; Google AI Ultra for Gemini 4 Carbon

Supported surfaces

Antigravity desktop App (Overview > Extensions > Swarm) and CLI (/swarm board or Alt+S)

Agent roles

Coordinator, named builders or pools, read-only @verifier (1 per 2 to 4 workers) and @monitor

Worker workspace modes

Shared git-worktree working copies, isolated repository branches or read-only inherited views

Lifecycle states

Active (working), Waiting (asleep on board at zero token cost) and Killed (permanent cleanup)

Release details & context

Worker lifecycle, quota hot-swapping and Swarm inspection views

Documented in the Google Antigravity /swarm blog and Swarm reference guide, coordinators spawn either role-named workers (such as @parser or @docs) or numbered pools (@fix-1 through @fix-4) over a shared append-only board. Extending Antigravity’s Remote Agents and preceding Antigravity Browser Use, waiting workers consume zero tokens until mentioned, and killed workers permanently delete temporary share or branch workspaces once changes are merged.

When a worker exhausts model quota, it pauses cleanly without auto-retrying on UI refresh until the developer messages it or hot-swaps that member to another model while retaining its full conversation history. Desktop users inspect runs across Board (with New dividers, search and mention filters), Members (Active, Standing by and Retired) and Graph timeline tabs, or press Alt+S in the CLI.

Common questions

What happens in Antigravity /swarm if a worker agent hits a model quota limit?

The worker pauses without burning retries on page refresh and can be resumed via a mention or hot-swapped in place to a different model while keeping its existing conversation history intact.

Do idle workers on the /swarm message board consume tokens while waiting?

No, workers in the Waiting state sleep on the shared message board at zero token cost until another member or the coordinator mentions them.

Google release timeline (24 launches tracked)

Earlier9 Nov 2026

Google unveils confidential TPU 8tx and high-memory TPU 8i Max with open-source TPU Sync

Google introduces confidential TPU 8tx and high-memory TPU 8i Max accelerators alongside Agent Compute Engine and the open-source TPU Sync KV-cache layer.

Earlier9 Nov 2026

Google AI Studio adds zero-setup Google Maps Platform toolkit with free daily sandbox calls

Google embeds a Google Maps Platform toolkit into AI Studio with 100 free daily sandbox calls and no credit card requirement.

Shipped next11 Nov 2026

Google launches Gemini Decisions API and Gemini Decision 1 at $0.10 per million input tokens

Google launches the Gemini Decisions API and Gemini Decision 1 at $0.10 per million input tokens and zero output charges, leading Google’s calibration suite.