# Choose your AI provider and model (/core/govern/ai-control-and-optimization/select-your-ai-provider)



<CalloutContainer type="info">
  <CalloutDescription>
    **Target Audience**: Organization admins and owners configuring AI for their team.
  </CalloutDescription>
</CalloutContainer>

### TL;DR [#tldr]

Each organization can choose which AI provider and model power its AI chat experience. By default, Toucan AI uses Anthropic's Claude Haiku 5.5. You can switch to another configured provider (or a specific model within it) from **Settings → AI behavior**, or leave it on **Default** to always follow Toucan's recommended provider and model.

***

### When to use this [#when-to-use-this]

Use this page when you want to run the AI assistant on a specific provider or model — for example, to meet a data-residency requirement, to standardize on a provider you already trust, or to try a more capable model for complex analysis.

The selection applies to your whole organization and covers the AI assistant end to end: the main chat, follow-up question suggestions, and conversation summarization.

***

### Who can change this [#who-can-change-this]

Changing the provider requires the **organization update** permission, which **member**, **admin**, and **owner** roles have. Users with the **explorer** role cannot change it.

***

### How to select a provider and model [#how-to-select-a-provider-and-model]

1. Open **Settings** and select **AI behavior** in the left sidebar.
2. In the **AI provider** card, open the **Provider** dropdown and pick a provider:
   * **Default (Anthropic)** — follow Toucan's recommended provider and model. Choose this if you don't have a specific preference; the model follows the default automatically.
   * A specific provider (for example **Anthropic**, **OpenAI**, **Mistral AI**, or **Scaleway**).
3. If you picked a specific provider, open the **Model** dropdown and choose a model, or leave it on **Default** to use that provider's recommended model.
4. Click **Save**. A confirmation (“AI provider updated”) appears once the change is applied.

<CalloutContainer type="info">
  <CalloutDescription>
    Only providers that your deployment has configured with credentials appear in the dropdown. If a provider you expect is missing, contact your administrator or Toucan support.
  </CalloutDescription>
</CalloutContainer>

***

### Available providers and models [#available-providers-and-models]

The exact list depends on which providers your deployment has configured. When available, the options are:

| Provider       | Models                                                                                                                                                                       |
| -------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Anthropic**  | Claude Opus 5.5, Claude Opus 5, Claude Opus 4.8, Claude Sonnet 5.5, Claude Sonnet 5, Claude Sonnet 4.6, Claude Haiku 5.5, Claude Haiku 4.5                                   |
| **OpenAI**     | GPT-4o, GPT OSS 120B                                                                                                                                                         |
| **Mistral AI** | Mistral Large 4, Mistral Medium 3.5, Mistral Small 4                                                                                                                         |
| **Scaleway**   | GLM 5.3, DeepSeek V4.1 Flash, DeepSeek V4 Flash, Mistral Medium 3.5 128B, Mistral Small 3.2 24B, Qwen 3.8 27B, Llama 3.3 70B Instruct, Gemma 4 26B (4B active), GPT OSS 120B |

***

### Good to know [#good-to-know]

* **Reset to default**: set the **Provider** back to **Default** and save to return to Toucan's recommended provider and model.
* **Scope**: the choice applies to every AI interaction in your organization, not to a single conversation.
* **Data handling**: switching providers changes which subprocessor receives your prompts and context. See [Third-party subprocessors](/core/govern/security-model/third-party-subprocessors) and [AI assistant data handling](/core/govern/security-model/ai-assistant-data-handling).

***

### Related pages [#related-pages]

* [AI FAQ](/faq-and-troubleshooting/ai)