Brain (LLM)

The Brain (LLM) is the model that reasons. It connects to the brain port of an AI Agent and defines which provider and model the agent uses to think and decide which tools to invoke.

Configuration

Field	Type	Required	Description
provider	select	Yes	LLM provider (see table).
model	aiModelSelector	Yes	Model, depending on the provider.
credentials_id	credentials	Yes	Provider credential.
attach_to	text	Yes*	Agent alias. *Filled automatically when connecting to the `brain` port.

Providers

provider	Service	Example model
openai	OpenAI (GPT)	`gpt-4o`
anthropic	Anthropic (Claude)	`claude-sonnet-5`
google	Google (Gemini)	`gemini-1.5-pro`
deepseek	DeepSeek	`deepseek-chat`
qwen	Alibaba Qwen	`qwen-plus`
groq / llama	Groq (Llama)	`llama-3.3-70b-versatile`
mistral	Mistral AI	`mistral-large-latest`
kimi	Moonshot (Kimi)	`kimi-k2-0711-preview`
openrouter	OpenRouter	`openai/gpt-4o`
together	Together AI	`meta-llama/Llama-3.3-70B-Instruct-Turbo`
local	Ollama / vLLM	`llama3.1`

Example 1 — OpenAI brain

Drag Agent: Brain (LLM) onto the canvas.
provider: openai, model: gpt-4o, credentials_id: your OpenAI credential.
Connect its output to the agent’s brain port (purple).

Example 2 — Local LLM (Ollama)

Ideal for on-premise, without sending data outside.

Create a local credential with your server’s base URL (e.g. http://localhost:11434/v1).
Brain — provider: local, model: llama3.1, credentials_id: the local credential.
Connect to the brain port.

The rest of the agent (tools, RAG, skills) works the same, whether it talks to OpenAI or your local model.