Gemini pricing 2026: Free vs Pro vs Ultra and API costs

What is Google Gemini?

At its core, Gemini is Google's family of AI models. It's "multimodal," which is just a fancy way of saying it works with more than text. It can process code, images, audio, and even video. If you want the full feature rundown, I've covered what Gemini AI is separately; here we're focused on the money. (And if you just need to switch it on or off, here's how to enable Gemini.)

To get a grip on the pricing, you first need to understand the tiers. The current lineup is built on Google Gemini 3, and each model is tuned for a different job, with its own price tag.

• Gemini 3 Pro: The top-of-the-line model, built for heavy, complex tasks that need real reasoning power, like advanced coding, deep analysis, or long-form writing.

• Gemini 3 Flash: The balanced, everyday workhorse. It's tuned for speed and cost-efficiency, making it a great all-rounder for most common business tasks.

• Gemini 3 Flash-Lite: The most budget-friendly and fastest model in the lineup, designed to handle a huge volume of tasks quickly without running up a massive bill.

You can get your hands on these models in three main ways: through the API for building your own apps, as part of a subscription for business or personal use, or through special developer tools. Each route prices differently, so let's dive into each one.

Gemini pricing for developers: The API model

For developers and businesses building custom applications on Gemini, the API offers the most control and power. It runs on a pay-as-you-go model, which is fantastic for scaling but can also lead to some surprisingly high bills if you're not keeping a close eye on things.

Understanding how token costs work in Gemini pricing

Before we jump into the numbers, we need to talk about "tokens." In the world of AI models, a token is basically a piece of a word. For example, the word "pricing" might be split into two tokens, "pric" and "ing." You get charged for the number of tokens you send to the model (your input) and the number it sends back (its output).

Gemini's API pricing is usually listed per one million tokens. To make things a little more confusing, the newer Gemini 3 models also bill "thinking tokens" as part of the output cost, which covers the model's background reasoning. That makes forecasting your budget with 100% accuracy a real challenge.

API pricing tiers

Here's a straightforward breakdown of the costs for the main API models so you can see how they stack up (per 1M tokens):

The older Gemini 2.5 Pro is also still around at $1.25/$10 if you want a cheaper reasoning model than the latest Pro tier.

Other factors affecting API costs

On top of those per-token rates, a few other features can sneak onto your final bill:

• Context caching: This lets you store and reuse parts of your prompts to save on input costs, but it comes with its own small storage fee.

• Grounding with Google Search: This lets the model pull in live information from the web. You get 5,000 free requests each month on Gemini 3, and after that it costs $14 for every 1,000 requests.

• Batch mode: If your tasks aren't urgent, you can process them in batches and get a 50% discount.

This level of complexity is a real challenge for teams that need predictable budgets. While developers can juggle these levers to optimize costs, support teams just need something that works without all the math. (I've put real numbers on that trade-off in my AI vs human agent cost breakdown.) Instead of forcing you to manage token consumption, eesel runs on transparent, pay-as-you-go pricing tied to the work it actually does. There are no per-seat fees, and you're never billed for the model's internal follow-ups within a resolution.

Gemini pricing: Subscription plans for business and individuals

If you're not building a custom app from scratch, Google's subscription plans are a much simpler way to get Gemini. These plans have a fixed cost per user each month, making them far easier to budget for.

Google Workspace plans

Google has folded Gemini features directly into its Google Workspace plans for businesses, so the AI is now included rather than a separate add-on. The idea is to give your internal teams an AI boost inside the tools they already use every day.

The pricing is billed per user, per month (these are the flexible, no-commitment rates):

• Business Starter: $8.40/user/month

• Business Standard: $16.80/user/month

• Business Plus: $26.40/user/month

Annual-commitment pricing is a bit lower (roughly $7, $14, and $22 respectively). Either way, your team gets Gemini help right inside Gmail, Docs, Sheets, and Meet for drafting emails, summarizing meetings, and creating presentations. I've broken the tiers down further in my Gemini Workspace pricing guide and the wider Google Workspace pricing post.

These are great for making internal employees more productive. But they're designed to assist your team, not automate their work. For customer-facing automation that actually resolves support tickets without an agent ever touching them, you need a different kind of tool. eesel plugs right into your helpdesk to provide an autonomous AI Agent, which a standard Workspace license just can't do.

For individuals: Google AI Pro and Ultra plans

For solo users, freelancers, or anyone who just wants a powerful AI assistant, Google sells a ladder of consumer plans:

• Free: $0. Access to Gemini 3 with daily usage limits.

• Google AI Plus: $4.99/month, with higher limits and 400 GB of storage.

• Google AI Pro: $19.99/month (the plan formerly called Google One AI Pro). You get the top-tier Gemini 3 Pro model, Deep Research and NotebookLM, 2 TB+ of Google Drive storage, and other Google One perks.

• Google AI Ultra: from $99.99/month, with the highest usage limits, 20 TB+ of storage, and early access to features like Deep Think. I've written a full Google AI Ultra breakdown if you're weighing the top tier.

It's worth noting these plans are tied to a personal Google account, so they don't have the admin controls, security features, or team collaboration tools that businesses rely on. They're a solid choice for personal use, but not really cut out for a team environment.

For developers and ops: Code Assist

For the more technical folks, Google has Gemini Code Assist, a separate subscription built to speed up the whole software development process.

• Standard Plan: $19/user/month (with an annual commitment)

• Enterprise Plan: $45/user/month (with an annual commitment)

This tool gives developers AI-powered code suggestions, chat help, and other features right inside their coding environment. The pricing is predictable, but it's an add-on cost focused squarely on developer productivity, not on powering the AI features in your final product.

A complete Gemini pricing comparison

With all these different options, it can still be tough to see which one is the right fit. This table puts all the Gemini pricing models in one place to help you decide.

If you're cross-shopping Gemini against other tools at this point, I've run the head-to-heads: Gemini vs ChatGPT, Gemini vs Perplexity, and Gemini vs Copilot are the most useful starting points.

As you can see, Google's pricing really makes you choose between a complicated, variable API model and a fixed-price license for internal tools. If budget is your deciding factor, I've also ranked the cheapest AI helpdesk apps. Neither Gemini route is built for automating frontline support. eesel fills that gap with a platform designed specifically for customer service and simple, predictable pricing. You can go live in minutes, not months, without needing a team of developers just to manage API costs.

How to choose the right Gemini pricing plan

Google's Gemini pricing is split up for different kinds of users, and the right choice really just boils down to your goal.

• Use the API if you're building a custom application from the ground up and need all the flexibility you can get.

• Choose Google Workspace to give your team's internal productivity a boost within the Google ecosystem. (My Gemini for Google Workspace guide goes deeper here.)

• Go for Google AI Pro or Ultra if you want a powerful personal AI assistant for yourself.

But for support, ITSM, and internal help desk teams, wrestling with Gemini's token math and setup is a major distraction from the real goal: solving problems for users faster. That's the exact problem I work on at eesel.

eesel plugs into your existing helpdesk, Zendesk, Freshdesk, Gmail, and more, pulls together all your knowledge sources, and deploys an autonomous AI agent that actually resolves tickets, not just drafts them. The pricing is transparent and pay-as-you-go: you pay for the work it does, with no per-seat fees and no surprise token bills. And before you commit a cent, our powerful simulation mode replays your historical tickets so you can see your exact resolution rate and ROI.

You can go live in minutes, not months. Try eesel for free and see just how simple and predictable AI for support can be.