The short version
- These are the models Spanner's in-app AI chat can use. Your organisation can use only the ones we've made available to it, and you choose among those for each conversation. Some have limited capacity, so we offer them only to particular organisations.
- Some models process your data in the EU. Others may process it in the United States, or wherever the provider has capacity.
- Every provider here is a paid business service that doesn't train its models on your data. Providers may keep it for a short time to detect misuse.
- Whether your data may go to a given model is your organisation's decision. This page gives you the facts to make it.
- We update this page before we add a provider or change where a model processes data.
The models
| Model | Made by | Run by | Where data is processed | Notes |
|---|---|---|---|---|
| Gemini 3.8 Flashgemini-3.8-flash | Google Cloud (Vertex AI) | EU | ||
| Gemini 4 Argongemini-4-argon | Google Cloud (Vertex AI) | Anywhere | ||
| Claude Sonnet 4.6eu.anthropic.claude-sonnet-4-6 | Anthropic | Amazon Web Services (Bedrock) | EU | Anthropic can't see your prompts or results. |
| Claude Opus 4.6eu.anthropic.claude-opus-4-6-v1 | Anthropic | Amazon Web Services (Bedrock) | EU | Anthropic can't see your prompts or results. |
| Kimi K3global.moonshotai.kimi-k3 | Moonshot AI, based in China | Amazon Web Services (Bedrock) | Anywhere | Moonshot AI can't see your prompts or results. |
| GPT-6.1 Terragpt-6.1-terra | OpenAI | Microsoft (Azure AI Foundry) | Anywhere | Limited capacity, so offered to particular organisations. Our Azure resource is in the United States, but Microsoft may process each request in any of its regions. |
| GPT-6.1 Solgpt-6.1-sol | OpenAI | Microsoft (Azure AI Foundry) | Anywhere | Limited capacity, so offered to particular organisations. Our Azure resource is in Sweden, but Microsoft may process each request in any of its regions. |
EU means the provider processes your data only in its EU regions. Anywhere means it may process your data in any of its regions worldwide, including the United States, depending on where it has capacity.
Whichever model you use, your request first passes through our AI service, which runs on Microsoft Azure. Our privacy policy explains what we keep and for how long, and our data processing terms cover any personal data your requests include.
Which models you can use
We make a set of these models available to each organisation. Spanner shows your users only those. If your organisation's policies need a different set, for example EU-only models, email enquiries@catmantoolkit.com and we'll change it.
Some models have limited capacity, so we offer them only to particular organisations. If we need to change the models available to your organisation, we'll tell you first, unless a provider withdraws or restricts a model without giving us time to.
How providers use your data
We use each provider's paid business service. Under their terms they don't use your prompts or results to train their models. They may keep them for a limited time to detect and prevent misuse, and to meet legal obligations. The terms below give the details.
Providers' terms and usage policies
When you use a model, its provider's usage policy applies to you, as the Spanner licence (EULA) requires. Using Spanner's AI chat explains more.
Microsoft (Azure AI Foundry)
Google (Vertex AI)
Amazon Web Services (Bedrock)
- AWS Data Privacy Center, including the AWS Data Processing Addendum
- Amazon Bedrock data protection, which explains that model makers can't see your prompts or results
- AWS Acceptable Use Policy
- AWS Responsible AI Policy
Anthropic (Claude models)
Copilot and other AI assistants
When you use Spanner from Microsoft 365 Copilot, Claude or another AI assistant, the assistant's own model does the work, under your organisation's agreement with its provider. The models on this page aren't involved, and no AI credits are used. Using Spanner with AI apps explains how that works.
Changes to this page
Models come and go. We update this page, and the date at the top, before we add a provider or change where a model processes data. To be told about these changes by email, write to enquiries@catmantoolkit.com.