Free tool · LLM economics
What will the tokens cost?
Every automation quote needs an API-cost line. Describe the workload once — tokens per run, runs per day — and see the monthly bill across every major model, side by side. The cheapest fit gets the trophy.
Rule of thumb: 1 word ≈ 1.3 tokens; a page of text ≈ 500–600 tokens. Cached input bills at ~10% of the input rate on Claude and OpenAI.
| Model | In $/M | Out $/M | Per run | Per day | Per month |
|---|
Prices as of July 2026 — refreshed weekly from provider pricing pages. Claude Sonnet 5 shows intro pricing (standard $3/$15 after Aug 31, 2026). The custom row takes any numbers.
Quote clients with confidence
"The AI costs roughly $40/month at your volume" wins deals that a shrug loses. Put the number in your proposal and the retainer.
Compare before you build
The same workload can differ 50× between models. Check whether the job actually needs a frontier model before you architect around one.
Caching changes the math
Repeated system prompts bill at ~10% when cached. Slide the cached share and watch high-volume workloads get dramatically cheaper.
Costing the job is step two. Finding the job is step one.
AgentScout finds AI-agent, n8n, and automation gigs, scores every one 1–10 with the pay decoded, and alerts you the hour they post. First 20 lock Pro at $8/mo for life.
More free tools: Scam Checker · Check Fit Lite · Bid ROI Calculator · Stack Cost Calculator · Proposal Openers · Starter Kits · Rate Index · Market Reports