We host on Hostinger — get up to 20% off your first plan.Get the link →
Referral link · we may earn a commissionReferral link — we may earn a commission at no extra cost to you.Download a working automation JSON
Can AI answer engines cite your site?
Outreach scripts that get replies
Find the 3 workflows costing you sleep
200+ prompts, ready to ship
12-month posting plan in 60 sec
Brand voice doc in 4 steps
Runway / Pika / Sora / Veo formats
Pick a model, enter your request volume and average token counts, and see an estimated monthly cost — plus how every other model stacks up at the same volume.
Pricing table as of August 2026 (verified against provider pricing pages). API pricing moves fast, so treat every number as a directional estimate, not an invoice.
Claude, GPT, and Gemini families.
Every model, same volume, ranked by cost.
One click, paste into a proposal or Slack.
6,000 requests/month at this volume · pricing as of August 2026 (verified against provider pricing pages)
$72.00/ month · Anthropic Claude Sonnet 5
Compare across providers, same volume
| Model | Tier | $/1M in | $/1M out | Est. monthly |
|---|---|---|---|---|
| OpenAI GPT-5.6-lunaFast/cheap tier for high-volume tasks. | fast/cheap | $0.2 | $1.2 | $5.40 |
| OpenAI GPT-5.4 nanoCheapest current-generation OpenAI tier. | fast/cheap | $0.2 | $1.25 | $5.55 |
| Google Gemini 3.5 Flash-LiteCheapest current-generation Gemini tier. | fast/cheap | $0.3 | $2.5 | $10.20 |
| Google Gemini 3.7 FlashPromotional rate through Dec 31, 2026; rises to $1.50 / $7.50 from Jan 1, 2027. | mid | $0.75 | $3.75 | $18.00 |
| Anthropic Claude Haiku 4.5Fast/cheap tier for high-volume, low-latency tasks. | fast/cheap | $1 | $5 | $24.00 |
| OpenAI GPT-5.6-terraMid tier for most production workloads. | mid | $2 | $12 | $54.00 |
| Google Gemini 3.1 Pro PreviewRates for prompts up to 200k tokens; longer prompts cost more ($4 in / $18 out). | flagship | $2 | $12 | $54.00 |
| Anthropic Claude Sonnet 5Standard price; a $2/$10 intro rate runs through Aug 31, 2026. | mid | $3 | $15 | $72.00 |
| Anthropic Claude Opus 5High-end reasoning tier below Fable 5. | flagship | $5 | $25 | $120 |
| OpenAI GPT-5.6-solOpenAI's current flagship text model. | flagship | $5 | $30 | $135 |
| Anthropic Claude Fable 5Anthropic's top-end model tier. | flagship | $10 | $50 | $240 |
All rates as of August 2026 (verified against provider pricing pages). This is not a live feed and provider pricing changes often — check the provider's own pricing page before budgeting.
Turn the numbers into a real system
Why this calculator exists
When I scope an AI automation build, the first question a client asks after "how long will it take" is "what's this going to cost me every month to run." Most people have never priced out an LLM API before — they know the $20/month consumer chat subscription, not the pay-per-token API rate that scales with usage.
Those two numbers are not related. A consumer subscription caps your usage; an API call bills you for every token in and out, every single request. A chatbot answering 200 customer questions a day on a flagship model can cost more per month than the subscription that inspired the project — or it can cost a few dollars if you pick the right tier. The difference is almost entirely model choice and prompt length, both of which you control.
This tool exists so you can run that math before you commit to a build, not after the first invoice surprises you. Pick a model, enter your real volume, and see where you land — then compare against a cheaper tier to see what accuracy you'd be trading for cost.
If the number that comes back changes your build plan — fewer tokens per call, a cheaper model for the bulk of requests, a flagship model only for the hard cases — that's exactly the kind of architecture decision I help clients make on a discovery call, before a single line of code ships.
Waseem, building from Bali · info@skynetjoe.com
Quick answers
Published provider pricing pages, condensed into a static table as of August 2026 (verified against provider pricing pages). This is not a live feed — no API is called and nothing is fetched from the providers. Rates change often, so treat every number here as an approximation and verify on the provider's own pricing page before budgeting real spend.
A few reasons: cached/prompt-caching discounts most providers offer, batch API discounts, volume-tier pricing at high scale, and the fact that real conversations vary in token count far more than a flat average captures. This calculator gives you a directionally correct monthly number, not an invoice-accurate one.
Roughly, 1 token is about 4 characters of English text. A typical chat message is 50-300 tokens; a long document or big system prompt can run into the thousands. If you're unsure, start with this tool's defaults and adjust until the volume matches what you actually see in your provider's usage dashboard.
Fully free, no email gate. Change the model and volume as many times as you want and copy the summary whenever you're ready.