# Requesty
Requesty is an [[AI Gateway]] and LLM router offering 600+ models across 30+ providers behind one OpenAI-compatible API. It sits in the same category as [[OpenRouter]], with routing, automatic failover chains, prompt caching, observability dashboards, role-based access control and PII masking.
## The honest pricing detail
Requesty charges a **5% markup on model costs**, with every feature included. That is worth pausing on, because most of the market went the other way: [[Vercel AI Gateway]], Cloudflare, [[Portkey]] and OpenRouter have pushed per-token markup to zero and monetize elsewhere. Requesty is charging for the layer openly instead of hiding the business model.
You can decide whether 5% buys you enough. For a team that would otherwise build failover chains and spend dashboards themselves, it often does. For high-volume traffic, 5% compounds fast.
## Data handling
GDPR compliance with a DPA on request, EU data residency options, and PII masking at the gateway. The residency option is the interesting one, since most US-centric routers ([[Ramp Router]] among them) simply do not offer Europe at all.
[[Bring Your Own Key (BYOK)]] is supported across providers, so you can keep your own contracts and still use the routing and observability.
## References
- Official website: https://requesty.ai/
- Documentation: https://docs.requesty.ai/
- Pricing: https://requesty.ai/pricing
## Related
- [[AI Gateway]]
- [[Model routing]]
- [[OpenRouter]]
- [[Vercel AI Gateway]]
- [[Portkey]]
- [[Ramp Router]]
- [[Bring Your Own Key (BYOK)]]
- [[AI Observability]]