Dust vs Gemini Enterprise Pricing and Agent Cost for Teams of 25 to 100

Dust is the cheaper purchase you can actually verify. A Pro seat is $24 a month on yearly billing, so 25 people cost $7,200 a year. Gemini Enterprise has no current public price, and the last starting price Google listed, $30 a seat for Standard, would put the same 25 people at $9,000 a year if it were a monthly rate.

That last condition matters. When Google did show prices, it never said whether $21 and $30 were annual commitments or month to month, and it never gave Plus or Frontline a figure of its own. Between October 5 and October 9, 2026, it removed the numbers entirely and retitled the page. So any Gemini Enterprise total you see, including ours, is a reference point built on a historical list price, not a quote.

Dust is cheaper on paper for a second reason too: its self-serve plan is narrow. The $24 seat sits inside a Business plan with 3 data connectors. If your team needs a fourth source, the honest Dust price is an Enterprise quote, and at that point you are comparing two quotes, not a quote and a price.

Dust Pro costs $7,200 a year at 25 seats, $14,400 at 50 and $28,800 at 100. At Google's last listed $30 Standard rate, Gemini Enterprise would be $9,000, $18,000 and $36,000. InSearch Business, with all seven connectors, is $9,600, $19,200 and $38,400.

Read the Gemini rows as an order of magnitude. They assume the old starting price was per seat per month and that you could buy it without a sales conversation. At 25 seats you can still try that: the Cloud console allows self-serve purchases up to 25 seats. At 50 or 100 you are past that cap and into a sales process or an invoiced account, where the price is whatever you negotiate.

Two Dust details move these numbers. Mixed teams are common, so a realistic Dust bill puts most people on Pro and a few heavy users on Max, five Max seats in a team of 50 adds $4,800 a year. And Dust has a setting that upgrades a member from Pro to Max automatically when they need more credits. Turned on without a budget owner watching it, that is a $96 a month jump per person.

Yes, in practice. Google launched Gemini Enterprise in October 2025 on the foundation of Google Agentspace, and it now sells those capabilities under the Gemini Enterprise name. That is why people still search "Dust vs Google Agentspace": they are comparing Dust with the product Google now sells as Gemini Enterprise.

The rename changed more than the logo. Gemini Enterprise is sold in editions, Business, Standard, Plus, Frontline and Pay-as-you-go, each with its own quotas, and it sits in Google Cloud rather than Google Workspace. Billing, seats and connectors are managed in the Cloud console. If your company is a Workspace customer with no Google Cloud footprint, expect setup work on the Cloud side before anyone can run a query.

For Dust, the comparison is the same either way. Dust is closer to what Agentspace was at launch, a place to build and share agents, while Gemini Enterprise has grown into a broad platform with video generation, notebooks, developer credits and Claude models available on pay-as-you-go terms.

Dust's self-serve Business plan allows 3 data connectors, 5 Spaces and 5 remote MCP servers, with more only on Enterprise. Google's Gemini Enterprise documentation lists 127 connectors, including Slack, Notion, Confluence, Jira, SharePoint Online, Box, Dropbox, GitHub, Salesforce, ServiceNow and Zendesk.

This is the row that ends most Dust evaluations at companies past about 30 people. A typical stack is a document store, a chat tool, a wiki and a ticketing system, which is already four. On Dust Business you choose three and leave the rest out, or you ask for an Enterprise quote.

Gemini Enterprise has the opposite problem. Breadth is there, but each connector is a data store you configure in a Google Cloud project, and Google documents quotas on those too: a default of 100 data stores per project, with a technical maximum of 500. That is plenty for most companies, but it is setup that someone in IT owns.

We are the narrow option here and we will say so. InSearch connects Google Drive, Slack and Notion on Team, and adds Confluence, Jira, Gmail and Salesforce on Business. That covers the core stack of most 25 to 100 person companies, not the long tail Gemini reaches.

Dust gives each seat a monthly credit allowance, 8,000 on Pro and 40,000 on Max, and charges 3 credits for any search of company data. Gemini Enterprise gives each seat a daily query allowance, 160 on Standard and 200 on Plus, but pools it across the edition, so individual users are not capped at their own share.

Google states the pooling rule directly: "Individual usage isn't capped at the per-user limit; all users in the edition draw from their edition's shared pool until it is depleted." Its own example is 10 Standard licenses sharing 1,600 queries a day. That is generous for a team with even usage. It is a problem when five power users and an automated agent drain the pool before the rest of the company logs in. Pools reset daily.

Dust's model fails differently. Credits are per seat and unused credits do not roll over, so a light user wastes most of their allowance while a heavy user runs out. On the Business plan there is no pay-as-you-go safety valve: Dust's documentation says workspace top-ups must be bought in advance, while its pricing FAQ suggests Pro and Max can continue through overage. Those two statements do not agree, so ask Dust which applies to your contract before you sign.

Overage on Gemini is clearer in one area: storage and data indexing above the pooled allowance costs $5 per GiB per month. Standard includes 30 GiB per user and Plus 75 GiB, pooled across the project.

Skip Dust if you need more than 3 data sources and do not want an Enterprise contract, or if your team is past 100 people. Skip Gemini Enterprise if you have no Google Cloud footprint, cannot buy through sales or an invoiced account, or need a firm price before the procurement process starts.

• Dust is a poor fit for a company whose knowledge sits in five or six systems, because the self-serve plan connects three. It is also a poor fit when nobody will own credit consumption, since auto-upgrades and unspent allowances both cost money quietly.
• Gemini Enterprise is a poor fit for a small team that wants to start this week with a card. Self-serve stops at 25 seats in the console, there is no published price, and Frontline requires 150 or more Standard or Plus users before it is available at all.
• Both are a heavier purchase than needed if what people actually want is to type a question and get one answer, with a citation, from the documents they already have. Agent builders and AI platforms carry features that job does not use.

Yes. If the real requirement is one written answer with citations across Drive, Slack, Notion, Confluence, Jira, Gmail and Salesforce, a dedicated search product does that job without agent credits or edition quotas. InSearch Business is $32 a seat a month billed annually, with every connector included.

InSearch searches each connected app, writes the answer, and links every claim back to the exact source document. Permissions come from each source and are checked at query time, so nobody sees a file they could not open in the original app. We do not train models on your content. And the price on our pricing page is the whole bill: no credit meter, no pooled quota, no change in rate when you pass 100 people.

That is less than either product on this page in one sense, since we do not build agents or connect 127 systems. It is more direct in another. If your team keeps asking "where is the latest version of this" and "what did we decide about that", start with a search tool and add an agent platform later if you find a job for it. Our unified search page shows what that looks like day to day.