AI Automation Pricing:
Complete 2024 Cost Guide

Workflow platform pricing: 2024 comparison

The Make.com operation cost deep-dive

Make.com's pricing is based on operations — each module execution in a scenario run counts as one operation. A scenario with trigger + AI call + JSON parse + Sheets log + Slack alert = 5 operations per item processed. At Make.com Core's 10,000 operations/month: this handles 2,000 items per month. At Make.com Pro's 40,000 operations: 8,000 items per month.

Operations exceed the limit: Make.com pauses your scenarios until the next billing cycle, or you can purchase additional operations at $9 per additional 10,000 operations. Monitoring your monthly operation usage mid-month and upgrading before hitting the limit is significantly cheaper than overage fees on some plans.

AI model API pricing: current rates and per-task costs

What "per million tokens" actually means for your budget

One token ≈ 4 characters of English text ≈ 0.75 words. A typical short email (subject + 200-word body): approximately 300–400 input tokens. At GPT-4o mini ($0.15/M input tokens): processing 1,000 emails for classification costs approximately $0.05 in input token fees plus output token fees for the classification JSON ($0.01–0.05 at $0.60/M output). Total: approximately $0.06–$0.10 per 1,000 emails.

At GPT-4o ($5.00/M input tokens): same 1,000 emails costs approximately $1.50–$2.50 — 25x more expensive for identical classification tasks. The first optimisation move for almost anyone: audit which automations use GPT-4o and migrate classification tasks to GPT-4o mini.

Realistic monthly cost calculator for common use cases

Cost optimisation: 6 moves that meaningfully reduce AI automation spend

1. Model tiering by task complexity

The single highest-impact cost optimisation: for any task that can be described as "pick the right option from a list" (classification, routing, simple extraction), use GPT-4o mini or Claude 3 Haiku rather than GPT-4o. Test both on 20 real examples. For most classification tasks, the quality difference is negligible while the cost difference is 20–33x. This one change typically reduces total API costs by 40–70% for mixed automation portfolios.

2. Input token reduction through preprocessing

Before passing content to the AI, strip it down to the minimum needed for the task. For email classification: the subject + first 150 characters of the body is usually sufficient — not the full 800-word email. For document extraction: extract only the relevant sections. A preprocessing step that reduces input from 1,000 tokens to 300 tokens saves 70% of the input token cost for that call.

3. Output format constraints

Longer AI responses cost more in output tokens. For classification tasks returning a single word or a short JSON object, add explicit length constraints to your prompt: "Return ONLY the JSON. Maximum 50 words for the assessment field." This prevents the AI from generating verbose responses when concise ones serve the task equally well.

4. Batching low-urgency items

Instead of processing each item individually as it arrives, batch low-urgency items and process them in a single scheduled run. A scenario that processes 50 newsletter articles once per week as a batch uses the same number of Make.com operations and API calls as processing them individually — but requires only one scenario execution overhead rather than 50. More significantly, batching allows you to use the Batch API (available for OpenAI) which provides 50% cheaper pricing for non-real-time processing.

5. Caching repeated inputs

If your automation regularly processes the same or similar inputs (the same FAQ questions, leads from the same companies, recurring document types), store previous AI outputs in a lookup table (a Google Sheet or database) and check the cache before calling the API. A cache hit costs nothing; only novel inputs require API calls. For automations with 20–30% repeat input rates, this can reduce API costs by a similar percentage.

6. Switch to n8n self-hosted for high-volume automations

At volumes where Make.com Core ($9/month, 10,000 ops) is consistently exhausted, n8n self-hosted on a $5/month VPS removes the per-operation ceiling entirely. The breakeven point: if you need more than Make.com Pro ($16/month, 40,000 ops), self-hosted n8n at $5/month VPS + the same API costs is cheaper. The tradeoff is the initial setup time (90 minutes) and ongoing maintenance responsibility.

Specialist API costs: transcription, enrichment, and document processing

Beyond the orchestration platform and LLM API, many production AI automation stacks include specialist APIs. These have separate costs worth planning for.

The most common unexpected cost in AI automation stacks is lead enrichment APIs. Clearbit and Apollo provide high-quality company and contact data but at prices that add meaningfully to total automation costs at volume. For teams with lower budgets, the LinkedIn native integration (via Make.com's LinkedIn module) and a web search step (via a search API) can approximate enrichment for significantly lower cost at the expense of some data quality.

Frequently asked questions