Skip to content
Built in Lahore, PakistanHow we check the math
mathbehind

How to Price an AI Feature So It Stays Profitable

Start from cost per active user, not cost per token. Here's how to estimate it, including chat history, and turn it into a price with a healthy margin.

By Muhammad Ahmad, Software EngineerLinkedIn· figures checked

Published , updated . 4 min read.

Want to run your own numbers?AI Feature Cost CalculatorOpen the calculator

Adding an AI chat or assistant to a product is easy. Knowing what it will cost per customer, and whether your price covers it, is harder. Token prices are quoted per million, which makes them look negligible until real usage arrives.

Think in cost per active user

Your customers pay per month, so your AI cost should be measured the same way: the average monthly cost of one active user. It depends on how many conversations they have, how long each one runs, and the model you use.

Don't forget chat history

In a normal chat, every new message re-sends the whole conversation so far. The input grows with every turn, so a 12-message conversation costs far more than two 6-message ones.

1,000 users, 10 conversations each, on Claude Sonnet 5
  1. Input tokens

    6 × 800 + 6 × 60 + 360 × 6 × 5 ÷ 2equals10,560

  2. Output tokens

    6 × 300equals1,800

  3. Per conversation

    10,560 ÷ 1M × $2.00 + 1,800 ÷ 1M × $10.00equals$0.0391

  4. Per month

    $0.0391 × 10,000 conversationsequals$391.20

That is $0.39 per active user per month. If the feature is part of a $10 plan, AI costs take about 4% of revenue, which is comfortable. If your heaviest users chat ten times as much, their cost is ten times higher, so look at your usage distribution, not just the average.

Turning cost into a price

  • Set a target gross margin for the feature, for example 80%. Price ≥ cost per user ÷ (1 − 0.80).
  • Put limits on the heaviest usage: message caps, a cheaper model for simple requests, or a higher tier for power users.
  • Recheck when models change. Prices fall regularly, and a model switch can halve your cost overnight.

Tip: Long system prompts and documents are re-sent with every message. Prompt caching, offered by most providers, can cut that repeated input cost sharply.

Three ways to charge for AI

  • Included with limits: the feature is part of a plan, with a monthly cap. Simple to sell, but you must set the cap from your cost data so heavy users cannot turn a customer unprofitable.
  • Credits: customers buy or receive a number of credits, and each AI action uses some. It ties price directly to cost and handles very different usage levels well.
  • A separate add-on: a flat monthly price per user for AI features. Easy to understand, but check that the heaviest users are still profitable at that price.

Price from margin, not from cost

Covering the model cost is not enough. If your business targets a gross margin of around 80%, which is typical for software, every dollar of AI cost needs about five dollars of price behind it. An AI feature that costs $1.20 per active user a month therefore needs roughly $6 of the price to be attributable to it. If that price does not feel right for the value delivered, the answer is usually a cheaper model, tighter limits or caching, rather than accepting a much lower margin.

Plan for the heavy users

AI usage is rarely spread evenly. It is common for a small group of users to generate most of the requests. Before launching, estimate what your top 10% of users will cost, not just the average. A fair-use limit, higher tiers for heavy use, or credits all protect you from a handful of accounts eating the margin of everyone else.

Model prices keep falling, so revisit

Model prices have been moving fast. In September 2026 alone, OpenAI released GPT-6 Sol at $2 / $10 and GPT-6 Luna at $0.10 / $0.50 per million tokens, about half the price of the models they replaced. A feature priced six months ago may now have much more margin than planned, or could move to a cheaper model with little loss in quality. Recheck the cost per active user every quarter and whenever you change models.

A worked pricing example

Using the AI Feature Cost Calculator, a chat feature for 5,000 active users, each with 10 conversations a month of 6 messages, costs about $1,956 a month on GPT-6 Sol, or about $0.39 per user. At an 80% target gross margin, that part of the product needs about $1.96 per user per month of price behind it. If the plan costs $20 a month, the AI feature takes about 2% of revenue at average usage, which is comfortable.

Now check the heavy users. If the top 10% have conversations three times as long, their cost can be many times the average, because input grows with every message. A fair-use limit, such as a monthly message allowance with a paid top-up, keeps those users profitable without affecting everyone else.

Communicating limits

Customers accept limits they can see. Show remaining usage in the product, warn before the limit is reached, and make upgrading one click. Hidden limits that suddenly block a user create support tickets and cancellations.

Questions people ask

How do I estimate the AI cost per user?
Measure the average monthly cost of one active user from conversations per month, conversation length and the model's token prices.
How should I price an AI feature?
Set a target gross margin and make the price at least cost per user ÷ (1 − margin), then limit the heaviest usage so outliers stay profitable.
Why do long chats cost more?
Every message resends the conversation so far, so input tokens grow with each turn.
How much should I charge for an AI feature?
Work out the AI cost per active user, then price for your target margin. At an 80% gross margin, $1.20 of monthly AI cost per user needs about $6 of price behind it.
How much does an AI chat feature cost per user?
In our example on GPT-6 Sol, about $0.39 per user a month for 10 conversations of 6 messages. Heavy users can cost many times the average.

Sources

  1. Claude Platform Docs: Pricing