LLM Providers¶

Pipelit supports multiple LLM providers through a unified credential system. Each provider is configured as a credential on the Credentials page and then selected in the AI Model sub-component.

Supported Providers¶

Credential Types¶

Anthropic¶

Create an Anthropic credential to use Claude models.

Required fields:

Optional fields:

Available models (auto-fetched or fallback list):

• claude-opus-4-6
• claude-sonnet-4-6
• claude-haiku-4-5-20251001
• claude-opus-4-5
• claude-sonnet-4-5

Setup:

1. Get your API key from console.anthropic.com
2. Go to Credentials page → Add Credential
3. Choose type LLM Provider
4. Set Provider to Anthropic, paste your API key
5. Click Test to verify connectivity
6. Click Save

flowchart LR
    Cred[Anthropic Credential<br/>sk-ant-...] --> AIModel[AI Model<br/>claude-sonnet-4-6]
    AIModel -.->|llm| Agent[Agent / Deep Agent]

OpenAI¶

Create an OpenAI credential to use GPT and o-series models.

Required fields:

Optional fields:

Setup:

1. Get your API key from platform.openai.com
2. Go to Credentials → Add Credential → LLM Provider
3. Set Provider to OpenAI Compatible
4. Set Base URL to https://api.openai.com/v1
5. Paste your API key → Test → Save

MiniMax¶

MiniMax uses the OpenAI-compatible API format.

Required fields:

Available models:

• MiniMax-M2.5
• MiniMax-M2.1

Setup:

1. Get your API key from platform.minimaxi.com
2. Create an LLM Provider credential with openai_compatible type
3. Set base URL to https://api.minimax.chat/v1
4. Use the Test button to verify and load the model list

GLM (Zhipu AI)¶

GLM models from Zhipu AI are OpenAI-compatible.

Required fields:

Available models:

• glm-4-plus
• glm-4-air
• glm-4-flash
• glm-z1-flash

OpenAI-Compatible APIs¶

Any API that implements the OpenAI /v1/chat/completions endpoint can be used. This includes:

For services that don't require authentication, set the API key to any non-empty string (e.g., not-needed).

Creating a Credential¶

1. Navigate to Credentials in the sidebar
2. Click Add Credential
3. Select type LLM Provider
4. Fill in the required fields for your provider
5. Click Test Credential — Pipelit will attempt a small API call to verify the key
6. Click Save

Using a Credential in a Workflow¶

1. Add an AI Model sub-component to your canvas
2. In the AI Model configuration panel, select your credential from the dropdown
3. Select the model from the list (populated from the credential's available models)
4. Connect the AI Model to an Agent or Deep Agent via the blue llm handle

Provider Capabilities¶

Custom Headers¶

For providers that require additional headers (e.g., API version headers, organization headers), add them to the credential's custom_headers field as a JSON object:

{
  "X-Custom-Header": "value",
  "anthropic-version": "2023-06-01"
}

Credential Security¶

All API keys are encrypted at rest using Fernet symmetric encryption. The API never returns the full key — responses show only the first 4 and last 4 characters with **** in between. See Security for details.