Skip to main content

​
Overview

The OpenAI integration provides access to ChatGPT models through the Chat completion endpoint. Supports all GPT models including GPT-4, GPT-4 Turbo, GPT-3.5 Turbo, and the latest o1 reasoning models.

​
Setup

1

Get API Key

Sign up at OpenAI Platform and generate an API key from the API keys section
2

Add Credential

In Flowise, navigate to Credentials and create a new OpenAI API credential with your key
3

Add to Canvas

Drag the ChatOpenAI node from the Chat Models category to your canvas

​
Configuration

​
Basic Parameters

​
credential
credential
required
Your OpenAI API credential containing the API key
​
modelName
string
default:"gpt-4o-mini"
The model to use. Available models are loaded dynamically from your account.Popular options:
  • gpt-4o - Latest GPT-4 Omni model
  • gpt-4o-mini - Faster, cost-effective GPT-4 variant
  • gpt-4-turbo - GPT-4 Turbo with vision
  • gpt-3.5-turbo - Fast and economical
  • o1-preview - Advanced reasoning model
​
temperature
number
Controls randomness. Lower values (0.1) make output more deterministic, higher values (0.9) more creative
​
streaming
boolean
default:true
Enable streaming responses for real-time output

​
Advanced Parameters

​
maxTokens
number
Maximum number of tokens to generate in the response
​
topP
number
Nucleus sampling parameter. Alternative to temperature for controlling randomness
​
frequencyPenalty
number
Penalize new tokens based on their frequency in the text so far (-2.0 to 2.0)
​
presencePenalty
number
Penalize new tokens based on whether they appear in the text so far (-2.0 to 2.0)
​
stopSequence
string
List of sequences where the API will stop generating. Separate multiple with commas
​
timeout
number
Request timeout in milliseconds
​
strictToolCalling
boolean
Enable strict mode for function calling to ensure JSON schema compliance

​
Vision Support

​
allowImageUploads
boolean
default:false
Enable image input for vision-capable models like GPT-4 Turbo and GPT-4o
​
imageResolution
string
default:"low"
Control image resolution for vision models:
  • low - Faster, lower cost
  • high - Better detail recognition
  • auto - Let the model decide

​
Reasoning Models (o1, o3)

Reasoning models like o1-preview and o3-mini have special parameters and don’t support temperature or stop sequences.
​
reasoning
boolean
default:false
Enable reasoning mode for o1/o3 models
​
reasoningEffort
string
Constrain reasoning effort:
  • low - Faster responses
  • medium - Balanced
  • high - Most thorough reasoning
​
reasoningSummary
string
Get a summary of the model’s reasoning process:
  • auto - Default behavior
  • concise - Brief summary
  • detailed - Full reasoning trace

​
Proxy & Custom Configuration

​
basePath
string
Custom API base URL for OpenAI-compatible endpoints
​
proxyUrl
string
HTTPS proxy URL for routing requests
​
baseOptions
json
Custom headers and configuration as JSON

​
Usage Examples

​
Basic Chat Model

// Configuration in Flowise UI
Model Name: gpt-4o-mini
Temperature: 0.7
Max Tokens: 2000
Streaming: true

​
Function Calling Agent

// Use with Tool Agent for function calling
Model Name: gpt-4o
Strict Tool Calling: true  // Ensures strict JSON schema
Temperature: 0.1  // Lower for more reliable function calls

​
Vision-Enabled Chat

// Enable image analysis
Model Name: gpt-4o
Allow Image Uploads: true
Image Resolution: high

​
Reasoning Model

// o1-preview configuration
Model Name: o1-preview
Reasoning: true
Reasoning Effort: high
Reasoning Summary: detailed
// Note: Temperature and stop sequences are automatically disabled

​
Best Practices

Model Selection

  • Use gpt-4o-mini for most tasks (cost-effective)
  • Use gpt-4o for complex reasoning
  • Use o1-preview for advanced problem-solving

Cost Optimization

  • Enable caching to reduce repeated calls
  • Set appropriate maxTokens limits
  • Use gpt-3.5-turbo for simple tasks

Function Calling

  • Enable strictToolCalling for reliability
  • Lower temperature (0.1-0.3) for tool use
  • Provide clear function descriptions

Performance

  • Enable streaming for better UX
  • Set reasonable timeout values
  • Use appropriate reasoning effort

​
Common Issues

OpenAI enforces rate limits based on your usage tier:
  • Add retry logic with exponential backoff
  • Monitor your usage in the OpenAI dashboard
  • Consider upgrading your usage tier
If you exceed the model’s context window:
  • Reduce maxTokens parameter
  • Implement conversation summarization
  • Use a model with larger context (e.g., gpt-4-turbo-128k)
o1 and o3 models have different requirements:
  • Don’t set temperature (automatically disabled)
  • Don’t use stop sequences
  • Enable reasoning parameter

Build docs developers (and LLMs) love

Get started for freeTalk to us