Intelligent routing now in beta

The best way to integrate any LLM
into your project.

The intelligent LLM gateway that grows with you. Smart routing, per-user quotas, and data residency — all through a single OpenAI-compatible API.

No credit card required • OpenAI-compatible

  • Pay as you go with auto top-up
  • Provider prices, zero margin on tokens
  • Smart routing across providers
  • Per-user limits and quotas built in
  • Enterprise-grade security & full privacy
  • Data & inference sovereignty (EU / US)

Drop-in replacement for OpenAI

Use our native SDKs, the Vercel AI SDK provider, or any OpenAI-compatible client.

// npm install @llmrelai/sdk
import Relai from "@llmrelai/sdk";

const relai = new Relai({ apiKey: process.env.RELAI_API_KEY });

const response = await relai.chat.completions.create({
  model: "anthropic/claude-sonnet-4",
  messages: [{ role: "user", content: "Hello!" }],
});

console.log(response.choices[0].message.content);
Built with RustUse from anywhere

Memory-safe, blazing-fast gateway. Call it from any language.

TypeScript logoTypeScript
Python logoPython
Go logoGo
Java logoJava
Rust logoRust

Built for production AI apps

Everything you need to ship AI features with confidence.

Smart Routing

Route by capability, cost, or latency. Use aliases like reasoning/cheapest and let Relai pick the best model for your needs.

Per-User Quotas

Set spending limits per end-user. Track usage and costs by user ID without building your own metering infrastructure.

Regional Residency

Keep EU data in Frankfurt, US data in Virginia. Automatic routing enforces data residency without code changes.

Pay list price. No markup.

We profit from scale, not by marking up your API costs. You pay standard Bedrock and Azure rates — we add no additional per-token fees.

$0
Platform fee
List price
Per-token cost

Your data, your region

Choose where your requests are processed. EU keys stay in Frankfurt, US keys stay in Virginia. Data residency enforced at the infrastructure level.

EU RegionFrankfurt

GDPR-compliant. Data never leaves EU boundaries.

US RegionVirginia

Low-latency for US workloads. Full model availability.

EU and US regions

Your data remains private

We route all inference through private cloud deployments on Bedrock and Azure. Your prompts and completions never touch Anthropic, OpenAI, or other AI company servers.

Private cloud inference

All requests route through Amazon Bedrock and Azure OpenAI — never to frontier lab servers. Your data stays in enterprise cloud boundaries.

Zero training on your data

Anthropic, OpenAI, and other AI labs never see or train on your prompts. Contractual guarantees from AWS and Microsoft.

No logging

Unlike public API endpoints, private cloud deployments don't log your conversations for model improvement or analysis.

Enterprise compliance

SOC2, HIPAA, and ISO 27001 certifications. DPAs with cloud providers — not frontier labs who may change policies.

Enterprise

Governance, compliance, and reliability
at the scale you need.

For teams running AI in regulated environments. Bring your security, compliance, and operations team — we'll meet them where they are.

Custom pricing • Annual contracts • Dedicated support

  • SSO, SAML & SCIM

    Centralized identity, role-based access, and automated provisioning for your whole org.

  • SOC 2, HIPAA, DPA

    Audit logs, signed BAAs, and data processing agreements ready for procurement.

  • Multi-cloud auto-failover

    Cross-provider routing with automatic failover so your AI stays up when a provider doesn't.

  • Dedicated regions & private deploys

    Run Relai in your VPC or in dedicated isolated regions with custom data residency.

  • BYOK & custom contracts

    Bring your own provider keys, negotiate volume discounts, and pay on annual invoices.

  • 99.95% SLA + named CSM

    Dedicated Slack channel, named customer success manager, and custom onboarding.

Start with $5 free credit

No credit card required. Get started in minutes with our OpenAI-compatible API.