I was building a small LLM-powered side project and forgot to account for my system prompt tokens. 800 tokens × 2,000 requests/day × input token rate = $340 invoice I wasn't expecting.

Built APICalculators to avoid doing this math in spreadsheets every time. It now covers 16 calculators: LLM APIs (GPT-4o, Claude, Gemini), vector databases (Pinecone, Qdrant, Supabase), serverless (Lambda, Cloudflare Workers), auth providers (Clerk, Auth0), payment processors (Stripe vs Paddle), and cloud VPS comparison.

Everything runs in the browser — no signup, no data sent anywhere.

Would love feedback on what's missing or wrong.