Loading page
About this tool
Work out the true AI cost of a support or sales conversation, including resent context, caching and human escalation.
The Cost Per Conversation Calculator prices a full multi-turn AI conversation, including the part most estimates miss: because chat APIs are stateless, every turn resends the system prompt and the transcript so far, so total input tokens grow as n*s + n*u + (u+a)*n(n-1)/2 rather than linearly with turns. It applies prompt-cache pricing to whatever share of input is cached, adds the expected cost of human escalation, and compares the all-in figure with handling the same conversation entirely by hand. Support, sales and product teams use it to set a defensible cost-per-contact number before committing to volume.
Open Cost Per Conversation Calculator on AltFTool — it loads instantly in your browser.
Enter the values you already know.
Fine-tune the options to match your scenario.
Read the result and use it in your planning or reporting.
Models the quadratic growth of resent context instead of multiplying turns by tokens.
Adds the expected human cost, which usually dwarfs the model cost at realistic escalation rates.
Splits input between fresh and cached rates so the saving shows up in the per-conversation figure.
Add the input and output token costs for every turn, remembering that each turn resends the system prompt plus the earlier transcript. For six turns with an 800-token system prompt, 120-token user messages and 260-token replies that is 11,220 input and 1,560 output tokens — at 3 and 15 per million, about 0.057 per conversation.
Because the transcript is resent on every turn, input tokens grow with the square of the turn count, not linearly. In the six-turn example above, 5,700 of the 11,220 input tokens — 51% of the input bill — are history being re-read, which is why summarising or truncating older turns is one of the highest-leverage cost fixes.
Yes, where the repeated prefix is large. Caching applies a lower rate to the unchanged portion of the prompt, so a long fixed system prompt and retrieved context benefit most; short prompts with highly variable content see little effect because there is nothing stable to cache.
On the model cost alone, usually by a wide margin — cents against the cost of several agent minutes. The economics turn on the escalation rate: at 20% escalation and seven agent minutes at 24 an hour, the expected human cost of 0.56 is roughly ten times the model cost, so reducing escalations matters far more than optimising prompts.
Add the Cost Per Conversation Calculatorwidget to your blog or website — free, responsive, no signup. Just keep the “Widget by AltFTool” credit link visible.
<iframe src="https://www.altftool.com/embed/widget/cost-per-conversation-calculator"
title="Cost Per Conversation Calculator — free AltFTool widget"
width="100%" height="640" style="border:0;border-radius:12px;overflow:hidden"
loading="lazy" referrerpolicy="no-referrer-when-downgrade"></iframe>
<p style="font-size:12px;margin:4px 0 0">Widget by <a href="https://www.altftool.com/tools/all/cost-per-conversation-calculator?utm_source=embed&utm_medium=widget">AltFTool — free online tools</a></p>