Prompts type

Technique Purpose Key Feature Use Cases
Zero-shot prompting Direct model response without examples. Relies on pre-trained knowledge. Simple Q&A, general tasks.
Few-shot prompting Guides model with a few examples. Improves contextual understanding. Sentiment analysis, classification.
Chain-of-thought prompting Encourages step-by-step reasoning. Solves logical, multi-step problems. Math, logical reasoning, data interpretation.
Instruction-based prompting Focuses the model on specific tasks. Clear commands for optimized responses. Task-specific outputs, content generation.
Role-based prompting Assigns a role to guide response tone/content. Shapes the response style. Teaching, professional content.
Contextual prompting Builds context progressively. Supplies background information for nuanced responses. Long-form generation, detailed tasks.
Meta prompting Guides the model to design/refine prompts. Prompts about creating prompts. Automation of prompt creation.
Self-consistency prompting Selects the most consistent output. Majority voting for reliability. Tasks with definitive answers.
Generated knowledge prompting Generates intermediate context before answering. Encourages preparatory thinking. Step-by-step guides, context-heavy tasks.
Dynamic prompt optimization Adjusts prompts in real-time. Refines outputs during interactions. Adaptive workflows, iterative tasks.
Automatic prompt engineering Programmatically generates/refines prompts. Reduces manual effort at scale. Large-scale applications.
Multi-prompt fusion Combines multiple prompts for balance. Addresses competing priorities in outputs. Creative + factual tasks.
Prompt chaining Structures interdependent steps. Output of one prompt becomes input for another. Document analysis, process workflows.
Directional stimulus prompting Guides tone/behavior with subtle cues. Provides influence without explicit commands. Style-guided generation, targeted writing.
Graph prompting Structures relationships in graph format. Helps interpret dependencies and hierarchies. Network analysis, relational data.