Why You’re Overpaying for Resume Tools
AI costs have dropped fast, but many job-search tools still charge premium prices. Here is what resume tailoring actually costs and why many tools are overpriced.
Here's a number worth knowing: running an LLM call to tailor a resume now costs roughly 0.01-0.07 kr depending on the model and length. Many resume tools still charge 200-350 kr a month. That gap is worth thinking about.
What changed with AI costs
Between 2022 and 2025, inference costs dropped by roughly 100x. Models that used to be expensive per call now cost tiny fractions of a krone. That's not a small shift - it fundamentally changes what's reasonable to charge for AI-assisted features.
Companies that built their pricing before this happened are now sitting on margin that isn't justified by their cost structure. The ones that built after have a genuine opportunity to charge less and still be profitable.
What you're actually paying for
When you pay around 275 kr/month for a resume tool, you're not paying 275 kr in AI costs. You're paying for the interface, the company's marketing spend (often enormous), support, and whatever margin they decided was acceptable.
None of this is inherently wrong - software has real costs beyond compute. But it does mean the AI component is a tiny fraction of what you're paying, which means comparing tools purely on "AI features" misses the point.
The tools that have aged poorly
Resume tools built around the original GPT-4 pricing are in an awkward spot. Their cost structure was calibrated for expensive compute. Now that compute is cheap, they haven't meaningfully changed what they offer or what they charge.
The tell: if a tool still talks about "credits" and makes you anxious about running out, they probably haven't updated their model to reflect current reality.
What fair looks like
A few things to look for: transparent pricing with no hidden credit systems, a free tier that's actually useful, features that go beyond AI text generation - job discovery, company research, application tracking. The AI tailoring should feel like one feature, not the whole product.
The market is correcting. But slowly.