$94.50 vs $95.24
what $100 of spend buys through each fee structure
Blog
OpenRouter vs Requesty: half a point apart, differently shaped
On paper it is 5.5% versus 5%. In practice the fees land on different things, one includes caching and EU residency, and the other includes the biggest catalog in the category.
Included
Requesty ships caching and EU data residency on every plan
400+
OpenRouter's catalog advantage; breadth is still its product
OpenRouter charges 5.5% on credit purchases; Requesty marks tokens up 5% instead, with caching included and EU data residency on every plan. $100 of spend buys about $94.50 of tokens through OpenRouter and $95.24 through Requesty. OpenRouter wins on catalog breadth; Requesty wins on built-in efficiency and EU requirements.
Where each fee lands
OpenRouter fees the credits: 5.5% on every non-crypto top-up, tokens at provider list prices, BYOK free to a million requests a month. Requesty fees the tokens: a 5% markup on usage, no credit-loading step in the math. The practical difference on $100 of spend is under a dollar; the structural difference is what you get for it.
Requesty includes caching on every plan, which matters more than half a fee point: cache reads bill at roughly a tenth of input rates, and our measured agentic session saved 2.6x from caching alone. If your workload reuses context and you were not going to wire caching yourself, the smaller vendor's default can beat the bigger vendor's fee.
Where each wins
OpenRouter: catalog breadth (400+ models versus a curated set), ecosystem momentum, and the BYOK free tier. If your job is comparing many models or chasing new releases the week they land, breadth is the product and OpenRouter owns it.
Requesty: the caching default, routing assists, and EU data residency on every plan, which is the clean answer to a question OpenRouter makes complicated. For EU-constrained teams the residency line alone can decide this comparison.
Every fee structure at your spend
OpenRouter 5.5% on credit top-ups; Requesty 5% token markup; Portkey $49/month flat; LiteLLM $0 licence with ops estimated at ~$1,000/month (two loaded engineer-hours a week) unless the box is ticked. Fees verified 2026-07-28. The crossover between 5.5% and $49 flat sits near $891/month.
What neither changes
Both are aggregator-shaped: you pick the model, you re-pick when the market moves, and provider variance on open-weights catalogs applies to both. The selection homework, worth 14x the fee on our measured workload, stays yours either way.
The category that removes that homework is the blended model: Pareto runs several LLMs per request, synthesizes one answer, and bills at cost. Receipts on the model card; competitor's disclosure applies.
Questions, answered
Is Requesty cheaper than OpenRouter?
Marginally, on pure fee math: $100 of spend buys ~$95.24 of tokens versus ~$94.50. The bigger swing is Requesty's included caching, worth far more than half a point on context-heavy workloads.
Does Requesty offer EU data residency?
Yes, on every plan, which is its cleanest structural advantage for EU-constrained teams.
Does OpenRouter have more models?
Yes; the 400+ catalog is the widest in the category and remains OpenRouter's core product advantage.
Which should agent builders pick?
Whichever ships caching into your stack fastest, because agent sessions reread context constantly; our measured session saved 2.6x from caching. Then read our agent-specific breakdown for the deeper meter.
Compare any two models
VS
Rates verified 2026-07-28. "Measured task" = our identical dashboard-generation prompt, metered where marked ✓ and list-math otherwise. Verbosity from the Verbosity Index, Edition 1. Data: prices.json.
Run the comparison on your own traffic. $100 to verify.
Buy Pareto API credits