Venice AI

Use Venice AI for chat, image generation, embeddings, and more

Actions9

Chat → Complete

AI-generated

Overview

This node integrates with Venice AI to perform chat completions using various AI models. It allows users to send a sequence of messages and receive AI-generated responses, supporting advanced options like streaming output, temperature control, token limits, and web search integration.

Common scenarios include:

  • Building conversational agents or chatbots that require dynamic AI-generated replies.
  • Automating customer support interactions with customizable AI personalities.
  • Generating creative text completions or continuing dialogues based on user input.
  • Enhancing chat responses with external web search data and citations for up-to-date information.

For example, you can provide a conversation history as an array of message objects and get a coherent AI response, optionally streamed in real-time for interactive applications.

Properties

Name Meaning
Model Model ID to use for the chat completion (e.g., venice-uncensored, llama-3.3-70b, qwen-2.5-vl)
Messages JSON array of message objects, each with role and content, representing the conversation history
Stream Boolean flag to enable streaming the response incrementally
Additional Fields Collection of optional parameters to customize the completion:
- Temperature Sampling temperature controlling randomness (0 to 2)
- Max Tokens Maximum number of tokens to generate in the completion
- Top P Nucleus sampling threshold (0 to 1)
- Frequency Penalty Penalizes new tokens based on their existing frequency (-2 to 2)
- Presence Penalty Penalizes new tokens based on whether they appear in the text so far (-2 to 2)
- Stop Sequences JSON array of sequences where the model will stop generating further tokens
- Seed Random seed for reproducibility
- Response Format JSON schema defining structured output format
- Tools JSON array defining tools or functions available during generation
- Character Slug Identifier for a Venice character persona to influence style (e.g., alan-watts)
- Enable Web Search Option to enable web search: Off, Auto, or On
- Enable Web Citations Boolean to include citations when web search is enabled
- Strip Thinking Response Boolean to remove "thinking" blocks from reasoning models
- Disable Thinking Boolean to disable internal reasoning/thinking features

Output

The node outputs a JSON object containing the AI completion result:

  • When streaming is disabled, the output JSON includes the full completion content.
  • When streaming is enabled, the output JSON contains:
    • content: The concatenated string of all streamed chunks.
    • chunks: An array of individual streamed response pieces, useful for incremental processing or debugging.

No binary data is produced by this operation.

Dependencies

  • Requires an API key credential for Venice AI service authentication.
  • Uses Venice AI REST endpoints for chat completions (/chat/completions).
  • Supports streaming via server-sent events or chunked HTTP responses.
  • No additional environment variables are explicitly required beyond the API credential.

Troubleshooting

  • Missing or invalid API credentials: Ensure the Venice AI API key is correctly configured in n8n credentials.
  • Malformed messages JSON: The Messages property must be valid JSON representing an array of message objects; invalid JSON will cause errors.
  • Streaming issues: If streaming is enabled but no data arrives, check network connectivity and API limits.
  • Unsupported model IDs: Using an incorrect or unavailable model ID may result in API errors.
  • Stop sequences JSON parsing: The Stop Sequences field expects valid JSON; malformed input will cause failures.
  • Web search options: Enabling web search requires proper configuration and may increase latency or require additional permissions.

Error messages typically reflect these issues and can be resolved by verifying inputs and credentials.

Links and References

Discussion