Skip to main content
PentAGI integrates with OpenAI’s advanced language models, including the latest reasoning-capable o-series models designed for complex analytical tasks.

​
Environment Variables

​
OPEN_AI_KEY
string
required
Your OpenAI API key. Obtain it from OpenAI Platform.
​
OPEN_AI_SERVER_URL
string
default:"https://api.openai.com/v1"
OpenAI API endpoint URL. Use the default for standard OpenAI API access.
​
PROXY_URL
string
Optional HTTP proxy URL for enhanced security and network isolation (e.g., http://your-proxy:8080).

​
Configuration Example

Add the following to your .env file: 
# Basic OpenAI setup
OPEN_AI_KEY=sk-proj-...
OPEN_AI_SERVER_URL=https://api.openai.com/v1

# Using with proxy for enhanced security
OPEN_AI_KEY=sk-proj-...
PROXY_URL=http://your-proxy:8080

​
Supported Models

OpenAI provider offers cutting-edge capabilities across multiple model families:

​
Reasoning Models (o-series)

Advanced models with step-by-step analytical thinking capabilities:
  • o4-mini - Cost-effective reasoning model for complex tasks
  • o3-mini - Balanced reasoning performance and cost
  • o3 - Flagship reasoning model for deep security analysis
  • o1 - Previous generation reasoning model

​
GPT-4.1 Series

Latest flagship models optimized for complex security research:
  • gpt-4.1 - Latest GPT-4 model with enhanced capabilities
  • gpt-4.1-mini - Cost-effective variant for high-volume tasks
  • gpt-5-mini - Next-generation mini model

​
Model Selection by Agent Type

PentAGI automatically selects appropriate models based on task complexity:
Agent TypeDefault ModelTemperatureMax TokensUse Case
Simple tasksgpt-4.1-mini0.53,000Quick queries and basic analysis
Simple JSONgpt-4.1-mini0.53,000Structured data extraction
Primary Agento3-mini / o4-mini-4,000Core penetration testing
Assistanto3-mini / o4-mini-6,000Multi-step security workflows
Generatoro3-mini / o3-8,192Report and exploit generation
Refinergpt-4.10.76,000Result refinement and analysis
Advisero3-mini / o4-mini-4,000Strategic recommendations
Reflectoro3-mini / o4-mini-3,000Analysis review and critique
Searchergpt-4.1-mini0.74,000Information gathering
Enrichergpt-4.1-mini0.74,000Data enrichment
Codergpt-4.10.26,000Exploit development
Installergpt-4.10.26,000Tool installation and setup
Pentestero3-mini / o4-mini-4,000Dedicated penetration testing
Reasoning models (o-series) use reasoning.effort parameter instead of temperature. The effort levels are: low, medium, and high.

​
Features

  • Advanced Reasoning: o-series models with step-by-step analytical thinking
  • Latest GPT-4.1 Series: Flagship models optimized for complex security research
  • Cost-Effective Options: From nano models for high-volume scanning to powerful reasoning models
  • Versatile Performance: Fast, intelligent models for multi-step security analysis
  • Proven Reliability: Industry-leading models with consistent performance

​
Pricing

Example pricing per million tokens (as configured in the provider):
ModelInput ($/M tokens)Output ($/M tokens)
gpt-4.1-mini$0.40$1.60
gpt-4.1$2.00$8.00
o3-mini / o4-mini$1.10$4.40
Pricing is subject to change. Check OpenAI Pricing for current rates.

​
Custom Provider Configuration

For advanced users, you can customize model selection per agent type using a YAML configuration file: 
simple:
  model: "gpt-4.1-mini"
  temperature: 0.5
  top_p: 0.5
  n: 1
  max_tokens: 3000
  price:
    input: 0.4
    output: 1.6

primary_agent:
  model: "o3-mini"
  n: 1
  max_tokens: 4000
  reasoning:
    effort: low
  price:
    input: 1.1
    output: 4.4
See Custom Providers for more details on YAML configuration structure.

​
Troubleshooting

​
Authentication Errors

If you see authentication errors:
  1. Verify your API key is valid at OpenAI Platform
  2. Ensure the key has sufficient credits
  3. Check for any typos in the .env file

​
Rate Limits

OpenAI enforces rate limits based on your account tier. If you encounter rate limit errors:
  • Upgrade your OpenAI account tier for higher limits
  • Implement retry logic in your workflows
  • Consider using different models for different tasks to spread load

​
Model Availability

Some models may require specific API access:
  • o-series models require access to the OpenAI reasoning API
  • GPT-5 models may require beta access
Check your OpenAI account settings for model availability.

Build docs developers (and LLMs) love

Get started for freeTalk to us