System Prompts

Browse system prompts organized by AI provider.

Explore official and leaked system prompts from Claude, GPT, Gemini, Grok, and other major AI models.

AI Providers

Anthropic

Anthropic models and system prompts

36 prompts available

Google

Google models and system prompts

11 prompts available

Misc

Misc models and system prompts

8 prompts available

Open-Canvas

Open-Canvas models and system prompts

4 prompts available

Open-Manus

Open-Manus models and system prompts

7 prompts available

OpenAI

OpenAI models and system prompts

57 prompts available

Perplexity

Perplexity models and system prompts

3 prompts available

Proton

Proton models and system prompts

2 prompts available

Uncategorized

Uncategorized models and system prompts

1 prompt available

xAI

xAI models and system prompts

6 prompts available

Tips for Using System Prompts

System prompts are most effective when they are specific, structured, and tested across multiple scenarios. Start with an existing prompt from a major provider like Anthropic, OpenAI, or Google as a baseline, then customize the persona, tone, output format, and constraints to match your use case.

  • Be specific about role and context. Tell the model exactly who it is, what it knows, and what kind of tasks it should prioritize.
  • Set explicit formatting rules. Specify whether responses should use bullet points, headers, code blocks, or plain prose.
  • Define boundaries clearly. Instruct the model on what topics to avoid, what fallback behavior to use, and how to handle ambiguous requests.
  • Test across models. The same system prompt may behave differently on GPT-5, Claude Sonnet, and Gemini. Use ZeroTwo to test across 60+ models in one place.

What Is a System Prompt?

A system prompt is a hidden set of instructions that AI companies inject at the beginning of every conversation with their models. Unlike user messages — which appear in the chat interface — system prompts operate silently in the background, shaping every response the AI generates. They define the model's persona, knowledge cutoff, formatting preferences, content restrictions, tool-calling behavior, and much more.

System prompts are the primary mechanism that transforms a general-purpose language model into a specialized product. The same underlying model — for example, GPT-5 or Claude Sonnet — can behave very differently depending on the system prompt applied. A customer service version might be instructed to stay on-topic and escalate edge cases. A coding assistant might be told to always explain its reasoning step by step. A research tool might be required to cite sources for every factual claim. For AI developers, studying real-world system prompts from Anthropic, OpenAI, Google, xAI, and Perplexity provides a direct window into prompt engineering best practices used at scale.