Last Updated: November 11, 2026
Summary: Google's AI agents come in three layers: the Gemini Enterprise app for employees (from $21 per seat per month for the Business edition), Gemini Enterprise Agent Platform for developers (the renamed Vertex AI, billed pay-as-you-go), and consumer agents such as Gemini Spark in the Gemini app. Developers build with the open-source Agent Development Kit (ADK) and connect agents with the A2A protocol, which Google donated to the Linux Foundation.
Google renamed most of its agent lineup in 2026, so many guides still cite old names. Vertex AI is now Gemini Enterprise Agent Platform, and Vertex AI Agent Engine is now Agent Runtime.
This guide maps each Google AI agent product, its cost from Google's official pricing pages (checked November 10, 2026), how it compares with Anthropic and OpenAI, and which option fits your company.
Free download: The AI Stats Sheet 2026. 36 key AI numbers for business on four printable pages, with the source under every figure. Join 2,000+ business leaders who read AI Business Weekly and get it in your inbox right away.
In this guide
What Are Google AI Agents?
Google AI agents are Gemini-powered systems that carry out multi-step tasks on a user's behalf, such as researching a topic, updating records across apps or answering customers. Google sells them through five main products, each aimed at a different buyer.
New to the concept? Start with our explainer on what AI agents are. Google's current lineup:
Product | Who it's for | What it does | Starting price |
|---|---|---|---|
Gemini Enterprise app | Employees and business teams | Use Google and partner agents, build no-code agents, connect company data | $21 per seat per month (Business edition) |
Gemini Enterprise Agent Platform | Developers and IT | Build, deploy, govern and monitor custom agents (formerly Vertex AI) | Pay-as-you-go, with a monthly free tier |
Managed Agents in the Gemini API | Developers | Run Google's hosted Antigravity agent in a sandbox with one API call (preview) | Model tokens and tools used |
CX Agent Studio | Customer service teams | Build voice and chat agents for customers | $0.50 per session |
Gemini Spark | Individuals | Personal agent that works across Gmail, Calendar, Drive and other Google apps | Google AI Pro ($19.99 per month) or Ultra |
What Happened to Vertex AI Agent Builder and Agentspace?
Vertex AI Agent Builder and Agentspace are older names. At Google Cloud Next on April 22, 2026, Google introduced Gemini Enterprise Agent Platform as "our new developer platform and evolution of Vertex AI," according to the Google Cloud announcement.
Google's official name changes page lists the mappings. The ones that matter most for agents:
Vertex AI Platform is now Gemini Enterprise Agent Platform.
Vertex AI Agent Engine is now Agent Runtime.
Vertex AI Studio is now Agent Studio, and Vertex AI Studio App Builder is now App Builder in Agent Studio.
Vertex AI Search is now Agent Search.
Agentspace no longer appears on Google's product pages. The old Agentspace product address now loads Google's "Gemini for business" page, which offers a 30-day free trial of Gemini Enterprise Plus.
What Is the Gemini Enterprise App?
The Gemini Enterprise app is Google's workplace product where employees "discover, create, share, and run AI agents in a single, secure environment," in Google's words. It is billed per seat, per month, and connects to company data in Google Workspace, Microsoft 365 and other business systems.

Gemini Enterprise puts research, no-code agents and partner agents in one workspace app for employees. Photo: Getty Images / Unsplash+
According to Google's Gemini Enterprise FAQ and edition pages, the app includes:
Deep Research agent: researches complex topics across internal and external sources and turns the findings into an easy-to-read report.
No-code agent building: Google's Business edition page calls this Workflow Builder, for automating repeatable tasks in plain language.
Agent Gallery and partner agents: Google says every partner agent in the gallery, from companies such as Salesforce, ServiceNow, Workday, Oracle and Adobe, has been validated for security and interoperability.
Connectors: Gmail, Drive, OneDrive, SharePoint, Jira, ServiceNow and more.
Google's edition comparison says administrators can enable Anthropic's Claude Opus 5.5 and Sonnet 5.5, billed as pay-as-you-go usage.
Gemini Enterprise editions compared
Edition | Users | Storage per user | Agent Marketplace | Full connector set | Enterprise security and compliance |
|---|---|---|---|---|---|
Business | 1 to 500 | 25 GiB, pooled | No | No (select connectors) | No |
Standard | 1 or more | 30 GiB, pooled | Yes | Yes | Yes |
Plus | 1 or more | 75 GiB, pooled | Yes | Yes | Yes |
Pay-as-you-go | 1 seat minimum to start | Pay-as-you-go | Yes | Yes | Yes |
Frontline | 150 or more Standard or Plus users | 2 GiB, pooled | Yes | Yes | Yes |
Source: Google Cloud's edition comparison, last updated November 6, 2026. Google's Business edition marketing page describes Business as built for teams of up to 300.
How Much Does Gemini Enterprise Cost?
Gemini Enterprise Business starts at $21 per seat per month, according to Google's Business edition page. Standard and Plus start at $30 per user per month, as reported by eWeek in August 2026; Google's own Standard and Plus pages now direct buyers to sales rather than showing a price.
Business, Standard and Plus each come with a 30-day free trial, according to Google's FAQ. The bigger differences are daily quotas. Google's quotas page lists these pooled per-user limits:
Quota (per user, pooled) | Standard | Plus |
|---|---|---|
Assistant queries per day | 160 | 200 |
Agents created per day | 1 | 10 |
Deep Research reports per day | 3 | 10 |
AI developer tool credit | $10 per month | $15 per month |
The Pay-as-you-go edition has no license fee and no quota caps. Storage and data indexing cost $5 per GiB per month, and other features bill at Agent Platform rates. Paid usage stops automatically at a project spending limit administrators set.
Worked example: seat costs
25-person team on Business: 25 × $21 = $525 per month at the starting price.
200 employees on Standard: 200 × $30 = $6,000 per month at the reported starting price.
Treat these as floors: quota overages and partner models bill on top.
What Is Gemini Enterprise Agent Platform?
Gemini Enterprise Agent Platform is Google's developer platform to build, scale, govern and optimize agents, and it replaced the Vertex AI brand in April 2026. It uses pay-as-you-go Google Cloud billing, separate from app seats.

Agent Platform hosts and governs agents on Google Cloud infrastructure, billed pay-as-you-go. Photo: Planet Volumes / Unsplash+
The platform launch post describes the main pieces:
Agent Studio: a low-code visual interface. Logic built there can be exported to ADK for full-code development.
Agent Runtime: hosting for agents, which Google says supports long-running agents that run autonomously "for days at a time," with sub-second cold starts.
Agent Registry: an index of every internal agent, tool and skill, so only approved assets reach users.
Agent Identity: a unique cryptographic ID for every agent, creating an audit trail.
Agent Gateway: secure connections between agents and tools, with Model Armor protection against prompt injection and data leakage.
Model Garden: more than 200 models, including Anthropic's Claude Opus, Sonnet and Haiku.
For companies comparing vendors, our roundup of AI agent platforms places Google's platform alongside Microsoft, AWS, Salesforce and others.
What Is Google's Agent Development Kit (ADK)?
Google's Agent Development Kit (ADK) is an open-source, code-first framework for building agents, available in five languages: Python, Go, TypeScript, Java and Kotlin. Google says more than six trillion tokens are processed monthly on Gemini models through ADK.

ADK is a code-first framework for developers who want full control over their agents. Photo: Behnam Norouzi / Unsplash+
A Google Cloud developer post from October 2026 says every ADK language includes a Model Context Protocol (MCP) client and supports A2A. The April 2026 upgrade added a graph-based framework for organizing agents into networks of sub-agents with defined logic.
ADK suits teams that write code and want full control. Our guide on how to build AI agents walks through the general process.
What Is the A2A Protocol?
The Agent2Agent (A2A) protocol is an open standard that lets AI agents built by different companies communicate and work together. The Linux Foundation formed the Agent2Agent project on June 23, 2025, seeded with Google's transfer of the specification, according to the Google Developers Blog.
A2A is no longer a Google-only bet. At its one-year mark in April 2026, the Linux Foundation reported that:
More than 150 organizations support A2A, up from more than 50 in April 2025.
Microsoft has integrated A2A into Azure AI Foundry and Copilot Studio, and AWS supports it through Amazon Bedrock AgentCore Runtime.
Version 1.0 is the project's first stable specification.
For buyers, A2A support reduces lock-in: a Google-built agent can hand work to an agent on another vendor's platform.
How Much Does It Cost to Run an Agent on Google Cloud?
Running an agent on Agent Runtime costs $0.085 per vCPU-hour and $0.009 per GiB-hour of memory after a monthly free tier, plus the model tokens the agent uses. Google's Agent Platform pricing page lists these list rates:
Resource | Free each month | List price after free tier | 3-year savings plan |
|---|---|---|---|
Agent Compute | First 50 vCPU-hours | $0.085 per vCPU-hour | $0.068 per vCPU-hour |
Agent Memory | First 100 GiB-hours | $0.009 per GiB-hour | $0.0072 per GiB-hour |
Agent Storage | First 1 GiB-month | About $0.30 per GiB-month | Lower with plan |
Google bills usage to the nearest second and does not charge Runtime for idle time while an agent waits for the next prompt between turns. Other agent services bill on the same units:
Memory Bank and Sessions (billing since September 1, 2026): $0.30 per GiB-month of storage, $0.085 per 3 million reads and $0.085 per 1 million writes.
Agent Gateway (billing since July 13, 2026): $0.085 covers 15,000 API calls or authorization requests.
New customers: eligible new Google Cloud users get $300 in free credits valid for 90 days.
Worked example: one internal agent
Take an agent using 1 vCPU and 2 GiB of memory that is actively working 200 hours in a month. Compute costs (200 minus 50 free hours) × $0.085 = $12.75. Memory costs (400 GiB-hours minus 100 free) × $0.009 = $2.70. Hosting comes to $15.45 for the month before model tokens.
Model tokens usually cost more than hosting. Our Gemini API pricing guide lists current per-token rates.
What Are Managed Agents in the Gemini API?
Managed Agents in the Gemini API let developers start a Google-hosted agent with a single call, without building any infrastructure. Google launched them in preview on May 19, 2026, powered by its Antigravity agent, which works inside an isolated Linux sandbox, according to the Google announcement.
Google's Antigravity agent documentation says pricing is based on the Gemini model tokens and tools the agent uses, on free and paid tier projects. Developers can set a token budget, and Google warns that autonomous multi-turn loops "can consume significant tokens."
What AI Agents Can Individuals Use in Gemini?
Google's main consumer agent is Gemini Spark, a personal agent that works in the background across Gmail, Calendar, Drive and other Google apps. Google's Gemini Spark page says it is available to Google AI Pro and Ultra subscribers over 18 in select countries, as well as select business users.
Google says Spark is designed to check with you before major actions, and its app connections are off by default. Other agent-style features include Deep Research, which is in the free Gemini plan, and Daily Brief, a morning digest that Google began rolling out to AI Plus, Pro and Ultra subscribers in the US in May 2026.
US prices on Google's subscriptions page: Google AI Plus is $4.99 per month, Google AI Pro is $19.99 per month, and Google AI Ultra is $99.99 or $199.99 per month.
Does Google Offer AI Agents for Customer Service?
Yes. Google's customer service agent builder is CX Agent Studio, part of Gemini Enterprise for Customer Experience, priced at $0.50 per session for both voice and chat, with limits applying, according to the CX Agent Studio page.
Google says CX Agent Studio supports human-like voices in more than 40 languages. At the listed rate, 10,000 customer sessions a month would cost 10,000 × $0.50 = $5,000. Our guide to voice AI agents compares other phone-agent options.
Free download: The AI Stats Sheet 2026. 36 key AI numbers for business on four printable pages, with the source under every figure. Join 2,000+ business leaders who read AI Business Weekly and get it in your inbox right away.
How Do Google AI Agents Compare With Anthropic and OpenAI?
All three companies now offer hosted agents billed mainly on model tokens. Google's difference is that it pairs them with a per-seat workplace app and its own cloud platform. Here is how the hosted developer offerings compare, based on each vendor's official pages:
Vendor | Hosted agent product | Status | How it is priced |
|---|---|---|---|
Managed Agents in the Gemini API (Antigravity agent) | Preview | Model tokens plus tools used | |
Agent Runtime on Gemini Enterprise Agent Platform | Available | $0.085 per vCPU-hour and $0.009 per GiB-hour after free tier, plus tokens | |
Anthropic | Claude Managed Agents | Beta | Token rates plus $0.08 for each hour a session runs |
OpenAI | Agents API (Codex harness) | Public beta | No additional fees; tokens and tools used |
Anthropic describes Claude Managed Agents as "a suite of composable APIs for building and deploying cloud-hosted agents at scale," with no platform or seat fees. OpenAI launched its Agents API in public beta on September 10, 2026, and its Agent Builder documentation says the visual Agent Builder is being deprecated and is scheduled to shut down on November 30, 2026.
Google also offers Anthropic's Claude models through Model Garden, so a company can use Google's governance and billing with Claude. Our guide to Claude AI agents covers Anthropic's side in detail.
Which Google AI Agent Option Should Your Business Choose?
The right option depends on who will build and use the agents:
Small team, no developers: Gemini Enterprise Business at $21 per seat, built for small teams (Google's edition docs cap it at 500 users).
Larger company with compliance needs: Gemini Enterprise Standard or Plus. Choose Plus if staff will build many agents or run heavy research, since it allows 10 new agents and 10 Deep Research reports per user per day versus 1 and 3 on Standard.
Uneven or experimental usage: the Pay-as-you-go edition, with a project spending cap set from day one.
Engineering team building custom agents: ADK plus Agent Runtime on Gemini Enterprise Agent Platform, starting with the $300 new-customer credit and the monthly free tier.
Fast developer prototype: Managed Agents in the Gemini API, with a token budget on every run.
Contact center: CX Agent Studio at $0.50 per session.
Several pieces, including Managed Agents and no-code agent publishing, are still labeled preview, so pilot one workflow before buying seats company-wide.
Sameer's take: I've tried Google Gemini and find it decent.
Frequently Asked Questions
What are Google AI agents?
Google AI agents are Gemini-powered systems that complete multi-step tasks for users. Google sells them through the Gemini Enterprise app, Gemini Enterprise Agent Platform, Managed Agents in the Gemini API, CX Agent Studio and Gemini Spark.
How much does Gemini Enterprise cost?
Gemini Enterprise Business starts at $21 per seat per month, according to Google. Standard and Plus start at $30 per user per month, as reported by eWeek, and a Pay-as-you-go edition has no license fee.
Is Vertex AI Agent Builder still available?
Google's agent-building tools now sit inside Gemini Enterprise Agent Platform, the new name for Vertex AI. Vertex AI Agent Engine is now Agent Runtime, and Vertex AI Studio is now Agent Studio, according to Google's official name changes page.
Is Google's Agent Development Kit free?
ADK is an open-source framework with public repositories in Python, Go, TypeScript, Java and Kotlin. You pay for the models and cloud resources your agents use, such as Agent Runtime compute and Gemini tokens.
What is Google's A2A protocol?
A2A, or Agent2Agent, is an open standard that lets AI agents from different vendors communicate. Google donated it to the Linux Foundation in June 2025, and more than 150 organizations supported it by April 2026, including Microsoft and AWS integrations.
The Bottom Line
Google now offers a complete agent stack: a per-seat app for employees, a pay-as-you-go developer platform, an open framework in ADK and an open protocol in A2A. Business teams should start with a Gemini Enterprise trial, while developers can build with ADK on the monthly free tier.
Confirm names and prices on Google's live pages before committing, since much of the lineup was renamed in 2026.
Related Articles
What Are AI Agents?: the basics, explained for business.
AI Agent Platforms: how the major vendors compare.
Claude AI Agents: Anthropic's agent tools for business.
How to Build AI Agents: a step-by-step guide.
Gemini API Pricing: token rates for every Gemini model.
Gemini AI Statistics: users, revenue and adoption.
⭐ Get more AI Business Weekly in Google: add us as a preferred source on Google.
Written by Sameer Khan with AI assistance, fact-checked against the sources linked above.
