In partnership with
Hello, hello! 👋🏻
We got some good data from OpenRouter as to what's actually happening with prompt lengths:
That 4x growth in average prompts?
Programming is doing the heavy lifting. Developers are feeding entire codebases into AI, averaging 25,000+ tokens per request.
That's 50+ pages of context per prompt 👀
Meanwhile, marketing prompts sit flat around 5,000 tokens. Same with finance, health, legal, and most other categories.
🤔 Good News and Bad News
The optimistic take: Marketers don't need to write novels. 5,000 tokens (~4 pages of context) is plenty for most marketing tasks.
The realistic take: Developers figured out that more context = better output. They're running circles around everyone else. Our marketing teams haven't caught up yet.
Easy setup, easy money
Making money from your content shouldn’t be complicated. With Google AdSense, it isn’t.
Automatic ad placement and optimization ensure the highest-paying, most relevant ads appear on your site. And it literally takes just seconds to set up.
That’s why WikiHow, the world’s most popular how-to site, keeps it simple with Google AdSense: “All you do is drop a little code on your website and Google AdSense immediately starts working.”
The TL;DR? You focus on creating. Google AdSense handles the rest.
Start earning the easy way with AdSense.
💻 The Developer Advantage
Programmers aren't writing longer prompts because they're verbose. They're:
Uploading entire files — style guides, existing code, documentation
Including examples — "Here's working code, make mine work like this"
Adding constraints — specific frameworks, error handling requirements, edge cases
The result? AI that performs like a junior developer who actually read the docs.
📣 What This Means for Marketers
You don't need 25,000 tokens. But you probably need more than you're using.
Instead of: "Write email copy for our sale"
Try: Upload your brand guide + 3 top-performing emails + specific offer details + audience segment info + a great ICP + what you want AI to avoid. That's maybe 8,000+ tokens—and it's the difference between "meh" output and "professional"
The developers aren't smarter. They're just giving AI more to work with.
Your turn.
Cheers,
Alec
P.S. For the data geeks, 5 other findings from the full OpenRouter report worth knowing:
Open Source Hit 30% — DeepSeek processed 14.37T tokens. "No budget" is no longer an excuse.
Reasoning Models Dominate (50%+) — People are using AI to think, not just search.
Programming Is 50%+ of Usage — Claude owns most of that market. The real ROI is in automation.
"Glass Slipper" Effect — First-to-solve creates lock-in. Users stick with what works (though DeepSeek shows a boomerang effect—users try alternatives, then come back).
Asia Doubled Usage — North America dropped to 47%, Asia hit 29%. Singapore alone is 9.2% of global usage.
P.P.S. for the AI sceptic in your life (I showed this to many family members over the holidays), OpenRouter Token growth by week (in Trillions) over 2025. The represent less that 1.7% share of AI tokens. We need to assume from the data points we have that Google, OpenAI, and Anthropic are growing in a similar way.
