Google’s best AI app is not Gemini, it is this VS Code substitute no person talks about

Google’s Antigravity is an AI-first coding surroundings constructed on the bones of VS Code. It will get much less consideration than Gemini, but it could be the corporate’s most attention-grabbing AI product. The key phrase right here is most attention-grabbing, although, not higher. Here’s what it does, the place it impresses, and the place it nonetheless frustrates.

What’s Google Antigravity?

An agent-first editor constructed on acquainted foundations

Antigravity is Google’s agent-first coding surroundings, launched on November 18, 2025, alongside the Gemini 3 mannequin household. On the floor it seems acquainted, as a result of it’s constructed on a fork of VS Code. Extensions, themes, and keybindings carry over, so anybody who has spent years in Microsoft’s editor can open it and really feel at residence inside minutes.

The distinction lies in what Google put on the heart of the expertise. Instead of treating AI as a sidebar that completes traces of code and fills in for you while you fail, Antigravity treats it as a set of autonomous brokers that may plan, write, take a look at, and debug software program with minimal supervision. While it is Google’s reply to each Claude Code and OpenAI Codex, its agent orchestration function makes it stand out from each.

The product’s lineage tells you numerous about the way it works. Google licensed expertise from Codeium (creators of the Windsurf editor) and introduced key members of its founding group into DeepMind. That explains why Antigravity seems like a detailed cousin of that device. Varun Mohan, Windsurf’s co-founder, now leads the undertaking from inside Google. The app runs on Windows, macOS, and Linux, and at launch it was free to obtain throughout a public preview, which instantly set it other than paid rivals similar to Cursor.

Antigravity will not be restricted to Gemini both. Users may work with Anthropic’s Claude fashions, which provides builders a selection of engine relying on the duty. That flexibility issues in a market the place no single mannequin wins each benchmark or fits each workflow. Meanwhile, instruments like Claude Code or Codex restrict you to fashions from a single developer.

Google itself frames the app as an experiment, and the label suits. Reviewers famous at launch that the corporate already provides different coding instruments, together with Jules and the Gemini CLI, and it was not apparent how they’d coexist. Still, the core concept is simple to know. You describe a objective, an agent builds a plan, and also you assessment the outcomes somewhat than typing each line your self. Whether that may be a good commerce depends upon how a lot you belief the output, and that’s the place the product’s design decisions begin to matter.

Where it shines

Parallel brokers, visible proof, and a browser that pitches in

The strongest argument for Antigravity is the way in which it reorganizes your workday. Alongside a regular editor view, it provides the Agent Manager, a devoted management heart the place a number of brokers can run directly, every dealing with its personal activity. One can refactor a module whereas one other writes checks and a 3rd checks the end in a browser. It resembles managing a small group greater than chatting with a single assistant, and for builders juggling a number of threads, that shift is genuinely helpful.

Artifacts are the opposite standout concept. Rather than dumping a wall of code and logs, brokers save plans, screenshots, and recordings of their browser periods so you may see what they did and why. Reviewing work visually is quicker than studying diffs line by line, and it helps people who find themselves much less comfy studying each line of generated code.

Browser integration closes the loop. Agents can open a web page, click on by means of it, and confirm {that a} change really behaves as meant, which spares you the fixed tab-switching between editor and preview. Review insurance policies add a measure of management, letting you determine whether or not the agent ought to pause and ask for approval or transfer forward by itself. Some early customers needed to modify that setting earlier than the habits felt proper, however the choice is there.

Then there may be the value. Google launched the device free throughout its public preview, with entry to Gemini 3 and different fashions, at a time when many opponents charged round twenty {dollars} a month. Even reviewers who stayed cautious in regards to the product’s route praised its contextual understanding of a codebase. Taken collectively, these strengths make Antigravity really feel much less like a VS Code clone with a chat window and extra like a distinct manner of constructing software program altogether.

Where it falls quick

Usage limits and reliability maintain getting in the way in which

The largest weak spot has little to do with design and every little thing to do with capability. During Google I/O in May 2026, the corporate launched compute-based utilization limits for its Gemini instruments, and Antigravity customers felt the change immediately. Some reported hitting their limits inside about an hour of labor.

Google responded by tripling the Gemini limits for Antigravity twice in a matter of days and resetting weekly quotas for paid plans, and Mohan acknowledged that individuals might exhaust their weekly allowance after solely a few work periods. Even after these will increase, customers identified that the boundaries remained decrease than they had been earlier than the brand new system arrived.

Reliability has been a recurring grievance as effectively. Developer boards in early 2026 crammed with studies of 503 “excessive visitors” errors and 429 rate-limit messages throughout free and paid tiers, and a few customers described ready days for quotas to return after quick bursts of use. Frustrated posts from paying subscribers described the device as almost unusable for stretches of per week, and Google’s help replies principally promised to cross the studies alongside internally.

There are smaller gaps too. The AI nonetheless wants pretty exact prompts, and its options deserve a handbook verify earlier than you merge something. Developers who depend on OpenAI’s fashions may also miss them, since these fashions and the Codex extension haven’t been obtainable contained in the app. Add the truth that Google calls the entire undertaking an experiment, and a cautious reader would possibly hesitate earlier than transferring a critical workflow over. The concepts are sturdy, however the basis beneath them nonetheless wobbles.

Antigravity is promising, however limits maintain it again

Antigravity reimagines the code editor round brokers, and its parallel workflows, visible artifacts, and browser testing are genuinely spectacular. Strict utilization limits and uneven reliability maintain it from being a day by day driver for everybody. It deserves consideration, however anticipate some friction.