syrus-gemma-4-26b

n8n community chat model sub-node for Syrus Gemma 4 26B, connecting directly into n8n AI Agent workflows.

Package Information

Released: this month
Downloads: 0 weekly / 0 monthly
Latest Version: 1.0.1

Available Nodes

Documentation generating…

Syrus Gemma 4 26B Chat Model icon
Syrus Gemma 4 26B Chat Model
Chat Model sub-node for Syrus Gemma 4 26B, connecting directly into n8n AI Agent workflows

Documentation

Syrus Gemma 4 26B Chat Model for n8n

Syrus AI Logo

Official community AI Chat Model sub-node for Syrus Gemma 4 26B on n8n.

Overview

  • npm package: n8n-nodes-syrus-gemma-4-26b
  • n8n display name: Syrus Gemma 4 26B Chat Model
  • Sub-node connection type: ai_languageModel (attaches directly to the Chat Model input of n8n's AI Agent node).
  • Service Endpoint: POST https://gemma-4-26b.syrus.ia.br/generate
  • Authentication: syrus-api-key header (securely managed via n8n credentials).
  • Repository: https://github.com/pablolira-1982/n8n-nodes-syrus-gemma-4-26b

Features

  • Genuine AI Language Model Sub-node: Appears directly under the AI Agent Chat Model connection picker in self-hosted n8n.
  • Multi-turn History: Maintains full conversation context mapping human and AI interactions to the Gemma contents array (role: 'user' and role: 'model').
  • System Instructions Support: Automatically translates system prompts into structured context guidelines for the model.
  • Granular Generation Controls: Independently configure maxOutputTokens, temperature, topP, topK, and stopSequences.
  • Zero Secret Exposure: Credentials are fully encrypted by n8n and isolated from execution parameters, outputs, and log streams.

Capability & Verification Matrix

Feature Support Status Notes
AI Agent Chat Model Connection Supported True ai_languageModel sub-node
Text Generation Supported Native POST /generate mapping
Multi-turn Chat Conversation Supported Preserves ordered message history
System Prompts Supported Formatted as instruction context
Stop Sequences Supported Passed as JSON string array to generationConfig
Streaming Not Supported / Unverified Syrus batch completion mode
Function / Tool Calling Implemented in adapter; requires Syrus endpoint support Sends Gemini-style functionDeclarations and maps functionCall/functionResponse; verify against the deployed Syrus API
n8n Self-Hosted Tested Node.js >= 20.x, n8n >= 1.0.0
n8n Cloud Unverified Community node sandbox restrictions may apply

Installation

Via n8n Community Nodes (Recommended)

  1. In your self-hosted n8n instance, navigate to Settings > Community Nodes.
  2. Select Install a community node.
  3. Enter the package name:
    n8n-nodes-syrus-gemma-4-26b
  4. Agree to the risks of installing third-party community code and click Install.

Manual Installation (Self-Hosted Docker or Bare Metal)

From within your n8n custom nodes directory:

npm install n8n-nodes-syrus-gemma-4-26b

Restart your n8n container or instance to load the node.

Credentials Setup

  1. In n8n, navigate to Credentials > New Credential.
  2. Search for Syrus Gemma API (syrusGemmaApi).
  3. Enter your private Syrus API Key.
  4. Save the credential.

Security Note: Never check in, log, or export workflows containing plain-text keys. The node injects the key into the HTTP transport header syrus-api-key during runtime execution.

Model Settings & Tuning Guide

Parameter Type Default Description & Recommended Values
Max Output Tokens Number 8192 Upper bound capacity for token generation. Use 512–2048 for concise responses, up to 8192 for long outputs.
Temperature Number 1.0 Controls randomness. Lower values (0.0–0.3) produce repeatable, structured answers. Higher values (0.7–1.0) encourage creativity.
Top P Number 0.95 Nucleus sampling threshold (cumulative probability mass). Keep between 0.8 and 0.95.
Top K Number 40 Limits next-token pool to the K most likely candidates. Standard starting point is 20–40.
Stop Sequences JSON Array [] Literal substrings where generation will halt immediately (e.g. ["END", "###"]).
Request Timeout Number (ms) 60000 Abort request if Syrus does not respond within this timeframe.

Stop Sequences Explained

Stop sequences must be entered as a valid JSON array of strings, for example:

["END", "###", "</answer>"]
  • Literal Matching: Stop sequences match exact character sequences, not regex patterns.
  • Do not use commas as plain text: Comma-separated strings are rejected to avoid splitting delimiters that contain commas.

Quickstart: Connecting to n8n AI Agent

  1. Add an AI Agent node to your workflow.
  2. In the canvas, find the bottom connection labeled Chat Model.
  3. Drag a connection and search for Syrus Gemma 4 26B Chat Model.
  4. Select your Syrus Gemma API credential.
  5. In the node options, adjust Temperature or Max Output Tokens if needed.
  6. Connect a trigger (such as Manual Trigger or Chat Trigger) and test!

An example exported workflow is provided in examples/agent-basic-workflow.json.

Troubleshooting

  • 401 / 403 Unauthorized: Verify that your Syrus API key is active and correctly pasted into the n8n credential.
  • 429 Rate Limit Exceeded: The upstream service has throttled requests. Reduce concurrent executions or introduce delay nodes.
  • 500 / 502 / 503 Server Error: Transient upstream issue. The node will retry with bounded backoff before reporting a failure.
  • Connection Timed Out: Check if your server has unrestricted outbound HTTPS access to https://gemma-4-26b.syrus.ia.br.
  • Node not visible in Chat Model picker: Verify that you are running a supported self-hosted version of n8n with LangChain integration enabled.

Development & Testing

# Clone and install dependencies
npm install

# Build TypeScript to dist/
npm run build

# Run unit tests
npm run test

# Dry-run package contents for npm
npm pack --dry-run

Tool-calling compatibility

The adapter exposes LangChain's bindTools() contract required by n8n's Tools Agent. It serializes tools using Gemini-style tools[].functionDeclarations[] and maps returned functionCall parts to LangChain tool calls. The deployed Syrus /generate service must accept this contract; a successful package build does not prove upstream tool-calling support. Test one real tool invocation before production use.

Changelog

1.0.1

  • Implemented the bindTools() Chat Model contract required by n8n Tools Agent.
  • Added Gemini-style tool declaration, function-call response, and tool-result mapping.
  • Added unit coverage for tool-calling message and response conversion.

License

MIT © 2026 Syrus AI.

Discussion