Is there a private ChatGPT for business?

"Private" gets used for three different setups. Each one answers a different question: who can read the prompts, who decides how long they are kept, and whose hardware runs the model. For a firm of 10 to 200 people, the right choice depends on which of those questions your clients, insurers, and regulators actually ask.

What does "private" mean for an AI chatbot?

In practice it means three separate things. No training: the vendor keeps your prompts and files out of the data it uses to improve its models. Retention control: someone at your company decides how long conversations are kept and can delete them. Containment: prompts are processed only by companies you have a contract with, or by no outside company at all. A business plan of a public chatbot gives you the first and part of the second. Self-hosting gives you all three, at a price.

How do the three kinds of private AI chat compare?

What do ChatGPT Business, Claude Team, and Copilot Chat each promise?

ChatGPT. OpenAI says data from ChatGPT Business, Enterprise, and its API is excluded from model training by default, and that you own your inputs and outputs. Business workspace admins can view, export, and delete members' conversations and control how long data is kept. Deleted conversations leave OpenAI's systems within 30 days unless the law requires longer. Business includes SAML SSO and MFA, connectors to Google Workspace, Slack, and Microsoft 365, and is sold for teams of 2 to 200 people.

Claude. Anthropic's commercial terms state that Anthropic "may not train models on Customer Content from Services." The Team plan adds SSO, central billing, and project sharing. Enterprise adds audit logs, role-based access, SCIM, and custom retention periods with a 30-day minimum. Without a custom period, Enterprise conversations stay until someone deletes them, and deleted chats leave Anthropic's back-end storage within 30 days.

Microsoft Copilot Chat. Staff who sign in with a Microsoft Entra work account on an eligible Microsoft 365 plan, including Business Basic, Standard, and Premium, get Copilot Chat at no extra cost. Their prompts and responses fall under Microsoft's Data Protection Addendum, stay out of foundation model training, and follow your retention, audit, and sensitivity label settings, with the exact controls depending on your plan. Two details matter. Web search queries go to Bing under separate terms, and an admin can turn web search off. And staff signed in with a personal Microsoft account get none of these protections.

Is ChatGPT Business private enough for a small business?

For most office work, yes, as long as everyone uses the company workspace. The protections belong to the workspace, so a client file pasted into someone's personal account falls under personal-plan terms. A business plan also leaves two things to you: what staff paste in, and which rules apply to client data. Is ChatGPT safe to use at work? lays out what each ChatGPT and Claude plan does with your data.

What is a self-hosted or on-premise LLM, and what does it cost?

A self-hosted model is an open-weights model downloaded and run on hardware you control, in your office or in your own cloud account. Prompts never reach an outside AI company. OpenAI itself publishes two open-weights models, gpt-oss-120b and gpt-oss-20b, under the Apache 2.0 license, and says the larger one runs on a single 80 GB GPU. For what "open" covers, see What is the difference between open weights and open source?.

The cost is mostly GPU memory. Hugging Face's rule of thumb is about 2 GB of GPU memory per billion parameters at standard 16-bit precision, so a 70-billion-parameter model needs about 140 GB just to load. Compressing the model to 8 or 4 bits cuts that, with some change in answer quality. On AWS, a g6e.12xlarge instance with four GPUs and 192 GB of GPU memory was listed at $10.49 an hour on demand in US East on October 7, 2026. Running it around the clock comes to about $7,660 a month by our arithmetic, before storage, backups, and the staff time to patch, monitor, and upgrade the model. Smaller models need less memory and cost less to run.

For most firms of 10 to 200 people, that arithmetic settles it. Self-hosting earns its cost when a contract or regulator requires prompts to stay on infrastructure you control.

How should a 10 to 200 person firm choose?

• Start from your client contracts. If a contract or regulator names where data may be processed, that requirement picks the option for you.
• Count what you already pay for. A firm on Microsoft 365 Business Standard already has Copilot Chat with enterprise data protection for staff who sign in with work accounts.
• Decide who sets retention. If clients ask how long their information sits in an AI tool, choose a plan where your own admin sets that period.
• Look for checks before the model. A company-administered app can check each prompt against your rules before it leaves and log who sent what, which a vendor's privacy terms alone leave to you.
• Price the upkeep. A self-hosted model needs someone to run it every week, and that salary belongs in the comparison.

To follow a prompt from the keyboard to the model and back, read Where does your data go when people use AI?.