Prompt Cost Calculator

A prompt cost calculator estimates how much it costs to run a specific prompt configuration at a given request volume using an LLM API. Unlike a simple token cost calculator, a prompt cost calculator lets you model the full per-request anatomy — separating the fixed system prompt, the variable user message, and the generated completion. By multiplying the per-request cost against your daily request volume, it projects your real API spend across daily, monthly, and yearly horizons. Our prompt cost calculator online runs entirely in your browser, making it private, instant, and completely free.

1. Choose a Use Case Preset (or go Custom): Select a preset scenario like Customer Chatbot, RAG Pipeline, or Bulk Classifier to auto-fill realistic token values — or enter your own system prompt, user message, and output token counts directly.
2. Set Your Request Volume & Model: Enter your requests per day and select the LLM model you are running on. The calculator immediately shows your cost per request, cost per day, and projected monthly and yearly totals.
3. Compare Across All Models: The cross-model table recalculates your exact prompt configuration against 13 models from OpenAI, Anthropic, Google, xAI, DeepSeek, Mistral, and more — sorted by lowest cost. Click any row to switch your active model instantly.
4. Model Growth: Add a monthly request growth rate to see how your API costs scale as your user base grows — useful for Series A pitches, runway planning, and SaaS pricing strategy work.

• System Prompt Size: System prompts are charged on every single request. A 1,000-token system prompt running at 10,000 requests/day adds up to 10M input tokens per day — a significant cost multiplier that many developers underestimate.
• Input vs. Output Pricing Gap: Output tokens are typically 3–6× more expensive than input tokens. Applications that generate long responses (emails, code, reports) have a dramatically different cost profile than short-output classifiers or yes/no decision agents.
• Request Volume Compounding: At 50,000 requests per day, even a tiny cost-per-request difference (e.g., $0.0001) adds up to $1,825/year. Choosing the right model tier for your quality needs can reduce costs by 5–20× at high volumes.
• Prompt Caching for Fixed Contexts: Most major providers (OpenAI, Anthropic, Google, xAI) offer prompt caching at 75–90% off for repeated static content. If your system prompt is large and constant, enabling caching is the single highest-leverage cost reduction available — not captured in base rates shown here.

The fastest way to reduce your prompt running cost is to trim your system prompt — every token in it is charged on every request. Use concise, structured instructions instead of verbose prose. Second, consider whether you need a frontier model for every request: models like GPT-4o mini, Gemini 2.5 Flash, Claude 3.5 Haiku, or Mistral Small cost 10–50× less than flagship models and perform excellently on structured, well-defined tasks. Third, enable prompt caching through your provider — it can eliminate 75–90% of the cost of repeated system prompt tokens at a stroke. Finally, for non-time-sensitive workloads, batch APIs from OpenAI, Anthropic, and xAI offer an automatic 50% discount with no quality trade-off.