cd /news/ai-tools/learn-how-antigravity-usage-intellig… · home topics ai-tools article
[ARTICLE · art-127495] src=dev.to ↗ pub= topic=ai-tools verified=true sentiment=↑ positive

Learn how Antigravity Usage Intelligence lets you monitor token consumption, cache savings, and quotas in Google’s Antigravity IDE.

An open-source VS Code extension called Antigravity Usage Intelligence gives developers visibility into token consumption, cache savings, and subscription quotas when using Google's Antigravity agentic coding IDE. The tool runs entirely locally, reading Antigravity session data and querying the local language server to break usage into fresh input, cache hits, output, and internal reasoning tokens, while showing per-model quota status and reset countdowns. It also adds a GitHub-style activity heatmap and a Session Inspector for auditing individual sessions.

by read5 min views3 publishedSep 12, 2026

Figure: Example Antigravity Usage Intelligence dashboard in VS Code, showing token usage charts and quotas.

For developers building with Google Antigravity (an AI coding IDE), tracking how many tokens, API calls, and reasoning steps each workflow uses is crucial. The open-source Antigravity Usage Intelligence VS Code extension exposes hidden usage details: per-session token breakdown (input vs cache vs output vs internal reasoning), live quota status, and a GitHub-style activity heatmap. This article explains how Antigravity usage is measured today, how you can monitor it, and how the extension makes it visible.

Google’s Antigravity is an agentic development platform (using Gemini/Claude models) that runs multi-turn AI tasks. Unlike a simple chat, an Antigravity session can generate millions of tokens per task. In fact, in active projects “pairing with frontier models pushes tens of millions of tokens daily”. Antigravity uses cached context, internal “thinking” tokens (especially with Claude/Gemini), and external tools, so one session’s workload can vary greatly.

Antigravity’s subscription plans impose token quotas: Pro and Ultra users get the highest quotas refreshed every 5 hours (with generous weekly limits), while non-Pro accounts receive a smaller quota refreshed weekly. However, the built-in UI mainly shows whether a request succeeded, not a breakdown of usage. Developers face issues like:

Antigravity Usage Intelligence solves this by running 100% locally and reading Antigravity’s session data. It connects to the local language server to fetch your account’s actual quota and then provides a real-time “command center” inside your IDE with the missing usage details.

The extension breaks down tokens into categories. The table below describes each:

Token Type What it Includes Example Source
Fresh Input Non-cached prompt text (new input) Your initial prompt or file contents
Cache Hits Tokens served from Antigravity’s cache Re-used conversation history or context
Output Model-generated response tokens AI’s answer text or code diff
Reasoning Internal “thinking” tokens (Claude/Gemini) Hidden multi-step reasoning tokens

Example (7-day run):

In this example, the vast majority of tokens came from the prompt cache (2.53B), highlighting how much work Antigravity saved by reusing context. These details only appear when you analyze the raw session data.

The extension also keeps track of your subscription quota. It uses Antigravity’s local language server RPC to query your plan details and usage. The dashboard shows, for each model family (e.g. Gemini Flash, Claude Sonnet): your tier (e.g. Google AI Pro), remaining quota percentage, used prompt/flow credits, and a live countdown to the next refresh. For example, you might see:

Gemini 3.8 Flash — Used: 73% (Reset in 02:15:42)
Claude Sonnet 4.6 — Used: 88% (Reset in 02:15:42)

By default, Pro/Ultra plans reset every 5 hours. If the language server is offline, the extension falls back to a rolling 5-hour window estimate. You can configure warning thresholds (e.g. 75%, 90%, 100%) to get alerted before you hit a hard limit.

Figure: Example weekly activity heatmap (each cell = one day; darker green means more tokens used).

The extension includes a GitHub-style heatmap showing your activity over days and weeks. Darker tiles mean higher token throughput. There is also a 24-hour chart that highlights your peak coding hours. For example, you might discover that most of your token usage happens on weekday mornings, with evenings much lighter. These visual cues help you spot patterns (heavy coding days, all-nighters, etc.) at a glance.

For detailed auditing, the Session Inspector pane lists individual sessions. For each session it shows: the unique session ID, project path, duration, turn count, and a token distribution bar (fresh vs cache vs output vs reasoning). It also lists what tools were used (e.g. run_command, view_file, grep_search, etc.) with counts. You can copy the session ID or export a markdown report. This makes it easy to investigate unusual cases: e.g. you might find one session used a large number of tokens because it ran a long grep_search on a huge codebase.

Ctrl+Shift+X), search for

   code --install-extension nirbhay-hiwse.antigravity-usage-intelligence
ovsx get nirbhay-hiwse.antigravity-usage-intelligence

.vsix from the

   code --install-extension antigravity-usage-intelligence-1.0.7.vsix

The extension relies on Python (no extra libraries). It auto-detects your Python installation. If needed, set antigravity-stats.pythonPath in settings to your python executable. After installation, it watches Antigravity’s local data and will automatically parse new sessions.

To use it, open the Command Palette (Ctrl+Shift+P or Cmd+Shift+P) and run Antigravity Stats: Open Full Screen Dashboard. This opens the interactive analytics view. You can also enable a daily token count in the status bar via the antigravity-stats.showStatusBar setting, and refresh data manually with Antigravity Stats: Refresh Stats.

Security: The extension uses read-only access (?mode=ro) to Antigravity’s local SQLite databases and connects only to 127.0.0.1 (no internet). No external telemetry is sent.

All timestamps and charts use your system’s local timezone. For example, if your computer is set to GMT+5:30, the heatmap’s “days” align with that day boundary, and countdown timers show local time. Currently, Google’s Antigravity quotas are the same globally, so region-specific settings aren’t needed. (Enterprise/managed accounts may have custom quotas, which the extension will display as reported by the local server.)

Antigravity Usage Intelligence brings transparency to AI coding workflows. By breaking down token usage (fresh vs cache vs reasoning), showing daily trends, and syncing with your official quota, it turns Antigravity from a black box into an observable system. This helps you optimize performance (e.g. maximizing cache usage) and avoid surprises from hidden limits. Install it and let your usage data guide your AI development.

Learn more and get the extension on GitHub. Happy coding!

Suggested cover images:

cover-dashboard.png – "Developer’s dashboard showing AI token usage and quotas" cover-heatmap.png – "Calendar heatmap of activity (each square = token usage per day)" cover-tokens.png – "Diagram of token types (fresh vs cache vs output vs reasoning)" cover-coding.png – "Person coding with AI assistant (illustrates AI-driven development)" cover-architecture.png – "Architecture diagram of Antigravity data flow (local storage, language server, extension)"

── more in #ai-tools 4 stories · sorted by recency
── more on @google 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/learn-how-antigravit…] indexed:0 read:5min 2026-09-12 ·