Setting Up AI Models

Available Models

WPiko Chatbot supports various OpenAI models, each with different capabilities and pricing:

Models

– Most advanced reasoning capabilities
– Extended reasoning range including “X-High”
– Supports: Reasoning Effort (None to X-High), Verbosity

– Next-generation reasoning model
– Default “None” reasoning for speed, capable of high depth
– Supports: Reasoning Effort (None to High), Verbosity

• GPT-5

– Complete reasoning and capability
– Best for complex, multi-turn and high-precision tasks
– Largest model in the GPT-5 family
– Deep reasoning
– Supports: Reasoning Effort (Minimal to High), Verbosity

• GPT-5 Mini

– Balanced performance and efficiency
– Ideal for general-purpose usage with reduced cost
– Moderate reasoning
– Supports: Reasoning Effort (Minimal to High), Verbosity

• GPT-5 Nano

– Ultra-fast, cost-effective
– Excellent for real-time or high-volume, straightforward queries
– Basic reasoning
– Supports: Reasoning Effort (Minimal to High), Verbosity

• GPT-4.1

– Full GPT-4.1 capability
– Best for complex tasks
– Highest quality responses
– Most resource-intensive

• GPT-4.1 Mini

– Balanced performance
– Good for most use cases
– Efficient resource usage
– Recommended for general chatbot operations

• GPT-4.1 Nano

– Efficient and cost-effective
– Fast response times
– Suitable for basic interactions
– Best for high-volume, straightforward queries

Model Settings

For GPT-5 and newer models, you can fine-tune the AI’s behavior with the following settings:

Reasoning Effort

Controls how deeply the model thinks before responding. Higher effort leads to better answers for complex tasks but increases response time.

• None (GPT-5.1/5.2 only): Fastest response, minimal thinking time.
• Minimal (GPT-5 family only): Very fast, light reasoning.
• Low: Balanced speed and thought.
• Medium: Standard reasoning depth (Default for GPT-5 family).
• High: Deep thinking for complex problems.
• X-High (GPT-5.2 only): Maximum reasoning depth for the most difficult queries.

Verbosity

Controls the length and detail of the response.

• Low: Concise, to-the-point answers.
• Medium: Balanced detail (Default).
• High: Comprehensive, detailed explanations.

Model Selection

Choosing the Right Model

• Navigate to AI Configuration
• Find the “Model” dropdown menu
• Select from available models based on your needs.
• If a supported model is selected (e.g., GPT-5 family), additional settings will appear:

– Reasoning Effort: Adjust based on problem complexity.
– Verbosity: Adjust based on desired response length.

• Consider the following when making your selection:

– Consider complexity of tasks
– Balance cost vs. performance
– Account for response time requirements

Performance Considerations

Cost Efficiency

• Consider your usage volume
• Monitor token consumption
• Balance quality vs. cost
• Start with lower-tier models and upgrade as needed

Best Practices

Model Selection Tips

• Test different models with your specific content
• Monitor performance and costs
• Adjust based on user feedback

Optimization Strategies

• Use the simplest model that meets your needs
• Regular performance reviews
• Monitor usage patterns
• Adjust model selection based on:

– Response quality
– Speed requirements
– Budget constraints
– User satisfaction

Troubleshooting

Common Issues

• Slow Responses

– Consider switching to a faster model
– Consider switching to a non reasoning model
– Check server resources

• Quality Issues

– Upgrade to a more capable model
– Review system instructions
– Test with different prompts