Cloud providers
This page explains how to configure each internet service provider in the AI chat pane. To run a model on your computer, see Configuring LM Studio or Configuring Ollama.
All providers on this page charge the API key holder for usage. Check each provider's pricing before starting.
Common steps
- Create a provider account and generate an API key. Each section links to its provider's website.
- In Digi3D.AI, select Window → AI chat.
- Select the provider in Provider.
- Click ⚙ to open Configuration under AI assistant. Expand the provider item, paste the key in API key and click Accept. Digi3D.AI requests the models available to your account.
- Select a model in Model.
- Enter an example request to check that the provider works.
Claude (Anthropic)
| Field | Value |
|---|---|
| API key | Create and manage it at platform.claude.com/settings/keys. |
| URL | https://api.anthropic.com/v1 (the default). |
| Model | The list suggests Opus 5.5, Sonnet 5, Fable 5.1 and Haiku 4.5, plus models returned by Anthropic for your account. |
Haiku 4.5 is the fastest, least expensive model and suffices for most requests. Larger models reason better through multistep requests at a higher cost.
OpenAI
| Field | Value |
|---|---|
| API key | Create and manage it at platform.openai.com/api-keys. |
| URL | https://api.openai.com/v1 (the default). |
| Model | The list shows your account's text models. Digi3D.AI excludes image, voice, transcription and embedding models because the assistant cannot use them. |
Azure OpenAI
Azure OpenAI serves OpenAI models from a Microsoft Azure subscription. It is commonly used by companies already using Azure services.
| Field | Value |
|---|---|
| API key | The Azure OpenAI resource key in the Azure portal: resource → Keys and Endpoint. |
| URL | The resource endpoint ending in /openai/v1. For example: https://mi-recurso.openai.azure.com/openai/v1. This field is initially empty and is required. |
| Model | The deployment name you created in Azure, rather than the model name. Select Other... and enter it. |
Azure does not allow Digi3D.AI to query your deployments, so no list is shown.
Kimi (Moonshot AI)
| Field | Value |
|---|---|
| API key | Create and manage it at platform.kimi.ai/console/api-keys. |
| URL | https://api.moonshot.ai/v1 (the default). |
| Model | The list suggests Kimi models, plus those returned by Moonshot AI for your account. |
OpenRouter
OpenRouter provides access to many vendors' models with one API key.
| Field | Value |
|---|---|
| API key | Create and manage it at openrouter.ai/settings/keys. |
| URL | https://openrouter.ai/api/v1 (the default). |
| Model | The list shows all OpenRouter models. Choose one supporting tool calling, which the assistant needs to work on the drawing. |
OpenAI-compatible server
The OpenAI-compatible provider supports any other server implementing the OpenAI API, such as a company server.
| Field | Value |
|---|---|
| API key | Whatever the server requires. Leave empty if not required. |
| URL | Server address, usually ending in /v1. For example: http://servidor:8000/v1. |
| Model | The list shows models returned by the server. If none appear, select Other... and enter the identifier. |
The model must support tool calling and a context of at least 32,768 tokens.
Troubleshooting
| Message or symptom | Cause and solution |
|---|---|
| No model selected for… | Model is empty. Select a model or enter its identifier using Other.... |
| Could not retrieve the model list… | Incorrect URL or API key, or no provider connection. Check with ⚙ (AI assistant in Configuration) and click ↻. |
| HTTP error 401 | Invalid or expired API key. Generate another on the provider's website. |
| HTTP error 429 | Account usage limit or balance exceeded. Digi3D.AI retries up to 3 times; if it persists, check the account on the provider's website. |
| The assistant replies with code instead of performing the task | The model does not support tool calling. Select another model. |