Google's greatest AI app isn't Gemini, it's this VS Code replacement nobody talks about

Published Oct 27, 2026, 5:00 PM EDT Arol is a tech journalist who currently works as a contributor at How-To Geek since 2022. He first began writing online for the short-lived portal of Spanish-language gaming forum Emudesc in 2013. Years later, in 2017, he got his true start in tech journalism working for a small Google-focused site called Pixel Spot. He transitioned to a news and feature writer role at XDA Developers that same year, where he worked until 2021 before making the jump to other websites. Arol brings nearly a decade of writing experience, and the occasional hot take, to his writings. While he's a technology lover at heart, he holds computer hardware and smartphones particularly close to heart. You'll normally find him covering news, although he has also written the occasional deal, buyer's guide, how-to post, and round-up. He's also written for Android Police and MakeUseOf. He's also a Political Science student at the Central University of Venezuela, and a proud member of student movement Àgora UCV. When he's not writing, you'll probably find him hitting the gym, trying to ace a new hobby, reading his textbooks, or traveling. You can reach him at me@arolwright.com. Google’s Antigravity is an AI-first coding environment built on the bones of VS Code. It gets less attention than Gemini, yet it may be the company’s most interesting AI product. The key word here is most interesting, though, not better. Here’s what it does, where it impresses, and where it still frustrates. What’s Google Antigravity? An agent-first editor built on familiar foundations Credit: Lucas Gouveia/How-To Geek Antigravity is Google’s agent-first coding environment, released on November 18, 2025, alongside the Gemini 3 model family. On the surface it looks familiar, because it is built on a fork of VS Code. Extensions, themes, and keybindings carry over, so anyone who has spent years in Microsoft’s editor can open it and feel at home within minutes. The difference lies in what Google put at the center of the experience. Instead of treating AI as a sidebar that completes lines of code and fills in for you when you fail, Antigravity treats it as a set of autonomous agents that can plan, write, test, and debug software with minimal supervision. While it's Google's answer to both Claude Code and OpenAI Codex, its agent orchestration feature makes it stand out from both. The product’s lineage tells you a lot about how it works. Google licensed technology from Codeium (creators of the Windsurf editor) and brought key members of its founding team into DeepMind. That explains why Antigravity feels like a close cousin of that tool. Varun Mohan, Windsurf’s co-founder, now leads the project from inside Google. The app runs on Windows, macOS, and Linux, and at launch it was free to download during a public preview, which immediately set it apart from paid rivals such as Cursor. Antigravity is not limited to Gemini either. Users can also work with Anthropic’s Claude models, which gives developers a choice of engine depending on the task. That flexibility matters in a market where no single model wins every benchmark or suits every workflow. Meanwhile, tools like Claude Code or Codex limit you to models from a single developer. Google itself frames the app as an experiment, and the label fits. Reviewers noted at launch that the company already offers other coding tools, including Jules and the Gemini CLI, and it was not obvious how they would coexist. Still, the core idea is easy to grasp. You describe a goal, an agent builds a plan, and you review the results rather than typing every line yourself. Whether that is a good trade depends on how much you trust the output, and that is where the product’s design choices start to matter. Where it shines Parallel agents, visual proof, and a browser that pitches in The strongest argument for Antigravity is the way it reorganizes your workday. Alongside a standard editor view, it offers the Agent Manager, a dedicated control center where several agents can run at once, each handling its own task. One can refactor a module while another writes tests and a third checks the result in a browser. It resembles managing a small team more than chatting with a single assistant, and for developers juggling several threads, that shift is genuinely useful. Artifacts are the other standout idea. Rather than dumping a wall of code and logs, agents save plans, screenshots, and recordings of their browser sessions so you can see what they did and why. Reviewing work visually is faster than reading diffs line by line, and it helps people who are less comfortable reading every line of generated code. Browser integration closes the loop. Agents can open a page, click through it, and verify that a change actually behaves as intended, which spares you the constant tab-switching between editor and preview. Review policies add a measure of control, letting you decide whether the agent should pause and ask for approval or move ahead on its own. Some early users had to adjust that setting before the behavior felt right, but the option is there. Then there is the price. Google launched the tool free during its public preview, with access to Gemini 3 and other models, at a time when many competitors charged around twenty dollars a month. Even reviewers who stayed cautious about the product’s direction praised its contextual understanding of a codebase. Taken together, those strengths make Antigravity feel less like a VS Code clone with a chat window and more like a different way of building software altogether. Where it falls short Usage limits and reliability keep getting in the way The biggest weakness has little to do with design and everything to do with capacity. During Google I/O in May 2026, the company introduced compute-based usage limits for its Gemini tools, and Antigravity users felt the change right away. Some reported hitting their limits within about an hour of work. 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 people could exhaust their weekly allowance after only a couple of work sessions. Even after those increases, users pointed out that the limits remained lower than they were before the new system arrived. Reliability has been a recurring complaint as well. Developer forums in early 2026 filled with reports of 503 “high traffic” errors and 429 rate-limit messages across free and paid tiers, and some users described waiting days for quotas to return after short bursts of use. Frustrated posts from paying subscribers described the tool as nearly unusable for stretches of a week, and Google’s support replies mostly promised to pass the reports along internally. There are smaller gaps too. The AI still needs fairly precise prompts, and its suggestions deserve a manual check before you merge anything. Developers who rely on OpenAI’s models will also miss them, since those models and the Codex extension have not been available inside the app. Add the fact that Google calls the whole project an experiment, and a cautious reader might hesitate before moving a serious workflow over. The ideas are strong, but the foundation underneath them still wobbles. Antigravity is promising, but limits hold it back Antigravity reimagines the code editor around agents, and its parallel workflows, visual artifacts, and browser testing are genuinely impressive. Strict usage limits and uneven reliability keep it from being a daily driver for everyone. It deserves attention, but expect some friction.