System Prompt Analyzer

Paste a system prompt and get structural analysis: token count, character count, heading detection, repetition warnings, and per-call cost estimates against current model pricing. Useful for compressing bloated prompts and verifying you're below cache-pricing thresholds.

What it analyzes

  • Token counts across GPT-5/4o (o200k_base), GPT-4/3.5 (cl100k_base), Claude, Gemini
  • Structural sections — counts H1/H2-style headings, bulleted instructions, numbered lists
  • Repetition — flags phrases (4-7 word n-grams) that appear 3+ times
  • Caching eligibility — whether prompt clears 1024-token threshold for Anthropic caching, 1024 for OpenAI
  • Per-call cost on each major model assuming the prompt is the full input