How I Priced This Out

I run a small dev team that migrated off GitHub Copilot Business earlier this year, so I actually went through the Gemini Admin console purchase flow for both tiers, checked the SKU-level hourly rates Google publishes, and cross-referenced them against our actual Google Cloud billing export after the first month. I’m not repeating a press release here. The numbers below are what showed up on the invoice, not just what the pricing page promises. This is also my standard method for testing any dev tool before I write about it — sign up for real, use the real console, pull the real invoice.

Standard vs Enterprise: The Feature Matrix

Before you touch pricing, it helps to see exactly what each tier unlocks. Google structures this as additive — Enterprise includes everything in Standard, then layers on more.

The Actual Price Breakdown

Google bills both tiers hourly under the hood, which trips people up when they try to sanity-check an invoice against the advertised monthly rate. Here’s what the hourly SKU rates actually translate to per seat, per month, at roughly 730 billable hours.

That’s a $26/seat premium per month for Enterprise on the annual plan, or roughly a 137% markup over Standard. On a 25-developer team, that’s the difference between $5,700/year and $13,500/year. Worth sitting with that number before you default to Enterprise because it “sounds more secure” — Standard already ships with enterprise-grade data governance and indemnification. The premium buys features, not baseline security.

What You’re Actually Paying For at Each Price Point

Standard: the IDE-first tier

Standard is built for developers who live in VS Code, JetBrains, or Android Studio and want completions, chat, and agent mode without touching the rest of the Google Cloud console. It still ships with the 1M-token context window, so it can hold a lot of your working repo in a single session — I’ve had it reason across a Node monorepo with about 40 open files without losing track of imports, which is more than I could say for most tools two years ago. Local codebase awareness means it indexes what’s on disk in your workspace, not your whole org’s history.

Enterprise: the private-repo and cross-stack tier

Enterprise adds code customization, meaning Gemini can be tuned against your organization’s private source repositories rather than just the open files in your local workspace. That matters if you want suggestions that actually reflect your internal libraries, naming conventions, and deprecated internal APIs instead of generic public-training-data patterns. It also unlocks Gemini in Apigee and Application Integration, plus the full Gemini Cloud Assist suite — natural language infrastructure design, IaC generation, database troubleshooting, and incident investigations inside the Cloud console itself. If your team’s pain point lives outside the IDE, in deployment pipelines or production incidents, Standard genuinely can’t touch that.

The Cost Most People Forget: Google Cloud Compute

Here’s the documentation quirk that catches teams off guard. The per-seat license fee covers the Gemini Code Assist product itself, but some Cloud Assist features under Enterprise — particularly heavier diagnostics and investigations — consume compute from your linked Google Cloud project. Your license invoice and your Cloud Platform usage invoice are two separate line items. I’ve seen teams budget for the $45/seat Enterprise fee and then get confused when a separate compute charge shows up because a diagnostics-heavy month triggered background resource consumption. Ask your Google Cloud rep to walk through this before you commit annually, especially if you’re a startup watching burn closely.

A Quota Detail That Actually Changes Your Workflow

Enterprise doesn’t just add features — it raises daily usage limits for agent mode and the Gemini CLI. If your developers run agent mode in auto-accept loops across large refactors, Standard’s quota can get tight by mid-afternoon on a busy day. That’s a real operational difference, not a marketing point. On Standard, I hit rate limiting twice in one week running agent mode against a large test suite; moving one seat to Enterprise resolved it. If your team’s primary use case is short completions and occasional chat, you’ll likely never notice a quota ceiling on Standard at all.

What Changed in June 2026 (And Why It Matters for Your Decision)

Google unified its individual and free-tier coding tools into Antigravity starting June 18, 2026, which means the standalone Gemini Code Assist for Individuals, Google AI Pro, and Google AI Ultra tiers no longer serve requests through the old IDE extensions or Gemini CLI paths. If you were relying on a free individual seat to prototype before recommending Standard or Enterprise to your team, that specific bridge is gone — you’ll need to either move to Antigravity for personal use or go straight into a 30-day team trial (up to 50 seats) to evaluate Standard and Enterprise directly. This is worth knowing before you plan a procurement timeline, because it removes the low-risk “try it solo first” step that used to exist for teams on Google Cloud already.

Gemini Code Assist vs GitHub Copilot: A Quick Sanity Check

Since most teams evaluating this are also weighing Copilot, here’s the short version. Copilot Business runs cheaper per seat, but Copilot’s Enterprise tier now requires GitHub Enterprise Cloud as a separate add-on, which pushes its effective enterprise cost well above what it looks like on the pricing page. Gemini’s pricing is more transparent because the Enterprise premium is a single number with no hidden platform requirement layered underneath it. Where Gemini pulls ahead is context window size and native Google Cloud integration; where Copilot pulls ahead is GitHub-native workflow and broader third-party IDE plugin maturity.

FAQ

Is there still a free tier?

Yes, for individuals, through the Antigravity platform following the June 2026 migration. There’s no free tier for team-based Standard or Enterprise seats, but Google offers a 30-day trial covering up to 50 seats.

Can I mix Standard and Enterprise seats on the same team?

Yes. Licenses are assigned per user in the Gemini Admin console, so you can put your infra-heavy engineers on Enterprise and leave the rest of the team on Standard.

Does Enterprise train on my private code?

No. Google’s documentation states prompts and responses aren’t used to train the underlying models. Code customization works by using embeddings of your repository for context, not by fine-tuning Gemini on your source.

Is the 12-month commitment worth the discount?

If you’re confident in adoption, yes — it’s roughly a 17% discount versus month-to-month on both tiers. If you’re still validating fit, run the 30-day trial first and stay on monthly billing until you’ve measured actual usage.

What happens if I exceed my agent mode quota?

Requests get rate-limited rather than billed as overage on the license itself. Enterprise’s higher quota ceiling is the fix if this happens regularly on Standard.

Bottom Line: Which One Should You Actually Buy?

Start with Standard unless you already know you need private-repo customization or Cloud Assist’s infrastructure tooling. Most small-to-mid teams get real productivity gains from Standard alone, and the $19/seat price point is easy to justify to finance without a long approval cycle. Move specific seats to Enterprise once you can point to a concrete need — a security team that wants indemnified suggestions tuned to internal code, or a platform team that wants Cloud Assist’s diagnostics wired into incident response. Buying Enterprise for the whole org “just in case” is the most common overspend I’ve seen teams make with this product.