LLM Nodes

The LLM nodes send a text prompt to a large language model and return the model's text response. Four provider nodes are included: OpenAI, Anthropic (Claude), Google Gemini, and Ollama. They share a common configuration and message contract so they can be used interchangeably in a flow.

Note: More providers, extra functionality, and multi-modality support for the LLM nodes are coming soon. Please reach out if you have such requirements.

Configuration

All four nodes share the same configuration fields:

• Model: The model name to call (for example, gpt-4.1-mini, claude-opus-4-6, gemini-3.1-flash-lite, llama3.2).
• System: An optional system prompt that sets the model's behaviour.
• Temperature: Sampling temperature controlling response randomness (default: 0.7).
• Max Tokens: The maximum number of tokens to generate in the response.
• Timeout: How long to wait for a response before aborting the request.
• API Key: Supplied via credential override or a provider environment variable. The Ollama node can also run against a local endpoint without a key.

API Key Environment Variables

Input

The node expects a text prompt in the configured input property (default: msg.payload). Non-string values are coerced to a string before being sent.

Output

• msg.payload - The model's text response.
• msg.ai_meta - Normalised response metadata, including provider, model, request_id, usage (input/output/total tokens), and rate_limits where the provider exposes them.

Note: The Ollama node returns its native response details on msg.ollama in addition to the normalised msg.ai_meta.

Example Flow

The following example sends the same prompt to OpenAI, Anthropic, and Gemini in parallel. It expects the relevant API keys to be available as environment variables (OPENAI_API_KEY, ANTHROPIC_API_KEY, and GEMINI_API_KEY). Each provider branch runs independently, so the flow works with only one or two keys configured.