Comparison
The five tiers of AI service, honestly compared
A recurring source of confusion is that "using ChatGPT" can mean five architecturally different things, each with different confidentiality, retention and accountability implications. This page walks through all five so you can identify which one your staff are actually using — usually not the one you assumed.
Credit where it's due
Where Free / Paid / Business / API cloud tiers genuinely wins
Every one of these tiers is a legitimate choice for someone. The question is not which is "best" — it is which matches your obligations, your team's actual usage patterns, and your appetite for depending on a third party. Most firms end up using more than one, whether by design or by drift.
Side by side
Green = private on-premise wins on this dimension. Black = Free / Paid / Business / API cloud tiers wins. Muted = it depends, or both have equivalent tradeoffs.
| Criterion | On-premise (OnPrem) | Free / Paid / Business / API cloud tiers |
|---|---|---|
| Free consumer accounts | The de facto tier for most staff using AI at home or on personal devices. Weakest defaults, longest retention, historically trained on inputs unless opted out. Zero visibility for the firm. | |
| Paid consumer subscriptions | Better than free — typically shorter retention, stronger defaults — but still fundamentally consumer terms. Personal account, no admin visibility. | |
| Business / Team / Enterprise tenancies | Real contractual commitments (no training, defined retention), SSO, admin controls, audit. A large step up. Protects only the usage that actually goes through the tenancy. | |
| API access | Different terms again — often no training by default, but usage patterns and data handling depend entirely on how the calling application is built. Not a "tier" for humans; a tier for developers. | |
| Private on-premise | Nothing transmitted anywhere. No vendor commitments needed because there is no vendor. Free at point of use. Constrained by hardware capacity and initial cost. |
Choose Free / Paid / Business / API cloud tiers if…
When the competitor is the right call
The right tier depends on the sensitivity of your work and what you can genuinely govern. Most firms with substantial confidential workloads end up wanting Business/Enterprise for general work AND on-premise for the sensitive core, with an explicit ban on free/paid consumer tiers for anything client-facing.
Choose on-premise if…
When we're the right answer
If you have never actually mapped which tier your staff use, that is the highest-value exercise on this list. Ask, without consequences attached — you will find at least one tier you did not think was in use.
Questions we get asked
- How do we find out which tier our staff are actually using?
- Ask them directly, without consequences attached. This works better than a survey and dramatically better than an audit. The behaviour is not shameful — the tool is genuinely useful — and staff will tell you honestly if the conversation is framed as "we're working out what to sanction" rather than "we're looking to discipline someone".
- Is API-access AI safer than a chatbot?
- It depends entirely on how the calling application is built and where it runs. An API-based tool built with proper data-handling controls can be very safe; a poorly built one can be worse than a consumer chatbot because it may retain and reuse data in ways nobody explicitly configured. The tier itself does not determine safety; the specific implementation does.
No obligation
Not sure which side you're on?
Tell us where you are today with Free / Paid / Business / API cloud tiers (or what's stopped you adopting it) and what work you actually need AI to do. We'll tell you honestly whether on-premise beats it for your situation — or whether you should just stay with what you've got.