Who does this task cheaper?
Pick a task, set the monthly volume. We price the person by the minute and the model by the token at published prices, add the human review, and show from what volume AI comes out cheaper.
Your task and volume
Task
Salary plus taxes, benefits and overhead divided by working hours. Not the net wage.
€4.67/ ticket
AI is cheaper from 246 tasks a month, with the build spread over a year.
€0.36/ ticket
Free 30-min call • No commitment
An estimate from your inputs and published prices. Real cost depends on prompt design, retries and how much review your process needs. · Currency switching uses fixed reference rates, not live FX
Same task, every model in the price list
Cost per task including the human review, for each model at its published price. The SaaS row is a per-resolution product price shown for scale; it is not a model you can pick.
Every number traces to an input or a published price
No black box and no industry factor. Below are the formulas with your current values substituted.
The task defaults are our working estimates from delivered projects, not benchmarks. Change any of them in the advanced mode and the whole page recalculates.
Retries, tool calls and system prompts ×1.2 · 1 USD = 0.92 EUR · 41.5 UAH
- 01Person per task = minutes / 60 × hourly cost8 min / 60 × €35= €4.67
- 02Model per task = (tokens in × price in + tokens out × price out) / 1M × retry factor(1,200 × 2 + 400 × 10) / 1M × 0.92 × 1.2= €0.0071
- 03Review per task = share checked × minutes / 60 × hourly cost20% × 3 min / 60 × €35= €0.35
- 04AI per task = model + review€0.0071 + €0.35= €0.36
- 05Difference per month = volume × (person − AI) − infrastructure4,000 × (€4.67 − €0.36) − €60= €17,178
- 06Saving in year one = difference × 12 − build€17,178 × 12 − €12,000= €194,141
- 07Payback = build / difference per month€12,000 / €17,178= 0.7 months
- 08Break-even volume = (build / 12 + infrastructure) / (person − AI)(€12,000 / 12 + €60) / (€4.67 − €0.36)= 246 / month
Published list prices, USD per million tokens
Prices as of 2026-08-27. Copied from the vendors' pricing pages on that date. Check the linked page before you sign anything.
| Model | Input | Output | Source |
|---|---|---|---|
| Claude Haiku 4.5Anthropic | $1 | $5 | Pricing page ↗ |
| Claude Sonnet 5Anthropic | $2 | $10 | Pricing page ↗ |
| Claude Opus 5Anthropic | $5 | $25 | Pricing page ↗ |
| Gemini 3.7 FlashGoogle | $0.75 | $3.75 | Pricing page ↗ |
| Gemini 3.1 ProGoogle | $2 | $12 | Pricing page ↗ |
| GPT-5.4 miniOpenAI | $0.75 | $4.5 | Pricing page ↗ |
| GPT-5.6 SolOpenAI | $4 | $20 | Pricing page ↗ |
| Intercom Fin (per resolution)SaaS, per resolution | $0.99 / task | Pricing page ↗ | |
What each task assumes
Minutes for a person, tokens for the model, how much of the output a human still checks and a representative one-time build in euros.
| Task | min | in / out | Review | Build |
|---|---|---|---|---|
| Support ticket reply | 8 | 1,200 / 400 | 20% × 3 | €12,000 |
| IT request (reset, access) | 12 | 900 / 300 | 15% × 2 | €15,000 |
| Lead research and qualification | 15 | 2,500 / 600 | 30% × 4 | €10,000 |
| Sales email or proposal draft | 40 | 3,000 / 1,200 | 100% × 8 | €10,000 |
| Invoice parsing and entry | 6 | 2,000 / 500 | 25% × 2 | €18,000 |
| Email classification and routing | 2 | 600 / 100 | 5% × 1 | €6,000 |
| RFP / security questionnaire answer | 25 | 4,000 / 800 | 60% × 6 | €14,000 |
| Payment reminder and collections touch | 6 | 1,500 / 300 | 30% × 2 | €14,000 |
| Your own task | 10 | 1,500 / 500 | 30% × 3 | €12,000 |
A&M Flow estimate from projects delivered in 2025 and 2026. Adjust to your own numbers.
Questions about the numbers
01How is the cost per task calculated?
The person's cost is minutes per task divided by 60 times the fully loaded hourly cost. The model's cost is input and output tokens times the vendor's published price per million, times 1.2 for retries and system prompts, plus the minutes a human spends checking the share of outputs you set. The formulas with your numbers are on this page.
02Where do the task defaults come from?
They are our working estimates from projects delivered in 2025 and 2026: how long a person takes, how many tokens a task uses and how much of the output still gets reviewed. They are not benchmarks. Every default is editable in the advanced mode.
03Why is there a human review line at all?
Because a model without review is not a process anyone signs off. The review share is the honest number in the model: at 20 percent one output in five is read by a person, at 100 percent every one is. Raise it when mistakes are expensive and the AI cost rises accordingly.
04What does break-even volume mean?
The monthly number of tasks from which AI is cheaper than the person once the one-time build is spread over a year and the monthly infrastructure is paid. Below that volume you are better off leaving the task with a person.
05Are the model prices current?
They are copied from the vendors' pricing pages on the date shown, with a link to each page. When the date is more than two months old the page tells you to check. Prices fall more often than they rise, so the estimate tends to be conservative.
06Can I get a tailored estimate?
Yes. Send us the link with your inputs or book a short call and we will pressure-test the minutes, tokens and review share against your actual process, then come back with a fixed scope and price.