Temperature
The Temperature slider controls how predictable or creative the AI’s responses are:- Reserved (0): Highly factual and predictable responses. Best for customer support, FAQs, or data extraction.
- Creative (>0): More varied and imaginative responses. Ideal for brainstorming, creative writing, or casual conversations.
Available Models
Each model offers unique strengths for different use cases:- GPT-4o 128K – Advanced, versatile model for complex and demanding tasks.
- GPT-4o-Mini 128K – Lightweight, fast, and cost-effective for simpler tasks.
- Claude Sonnet 4 200K – Balanced model with strong reasoning for multi-step problem solving.
- Claude 3.5 Haiku 200K – Extremely fast and affordable, best for high-volume and lightweight tasks.
- Gemini 2.5 Flash 1M – Efficient blend of reasoning and speed, adaptable to a wide range of tasks.
Custom Reasoning
With Custom Reasoning, the AI can process complex, multi-step problems more effectively.- Available only for: Claude Sonnet 4 200K and Gemini 2.5 Flash 1M.
When Custom Reasoning is enabled, you can adjust the Max Reasoning Tokens setting.
- This determines the AI’s “thinking budget” before generating a response.
- Higher token count: Better quality answers for complex queries (but may increase latency).
- Lower token count: Faster responses, suitable for straightforward tasks.