One OpenAI-compatible gateway to 250+ models from 40+ providers, with retries, failover across providers that serve the same model, spend limits and per-request costs built in. Enterprise adds SAML SSO, audit logs and guardrails, in our cloud or yours.
30-day production pilot
Your traffic, your security review, our engineers next to you.
Tell us about your stack, then pick a time with our team.
Enterprise control plane
Give every team the models they need without handing out provider keys. Identity, data rules, spend and reliability live in one place, with admin changes recorded in an audit log on Enterprise.
Sign in with Okta, Entra ID or another SAML 2.0 provider, with SCIM provisioning. Roles per organization and project, and API keys scoped per app.
Detect emails, card numbers, secrets, prompt injection and jailbreak attempts in chat requests. Block, redact or warn, per organization or project.
Set a spend cap on any API key or member and requests stop once it is reached, with optional alerts before a key runs out. Team budgets on Enterprise.
When a provider errors, times out or rate-limits you, the gateway retries up to twice on another provider for that model before it responds. Requests pinned to one provider stay there.
API key, invite, role, budget and compliance policy changes, with who made them and when.
06 / Deployment
Same gateway and dashboard, in our cloud or yours. The core is open source under AGPLv3 and the enterprise code is source-available, so nothing is a black box.
Enterprise Cloud
We run it for you with a 99.9% SLA and dedicated support.
Self-hosted
Docker or Kubernetes with our Helm chart, inside your own network. Enterprise features need a license key.
Provider policy
On Enterprise, route only to providers that meet your rules on headquarters country, data retention and training.
The LLM Gateway family
The same routing, billing and trust layer powers every product. Pick the gate for the people you're equipping.
One OpenAI-compatible API for 250+ models, with routing, caching, failover and cost analytics on every request. Guardrails on Enterprise.
Cost, latency, errors and cache hits on every request, with spend by model, provider and API key, and by project on Enterprise. Full prompts and responses when you turn on data retention.
Flat-price plans for AI coding. One key for your coding tools, included model usage, and every request tracked by tool in one dashboard.
Chat with GPT, Claude and Gemini, create images, video and speech, and run group chats with several models. Fast models from $9/mo, flagship models from $19/mo.
List your models on LLM Gateway, file your prices, pass review and compete for traffic from teams across the network.
For developers
Point any OpenAI SDK at LLM Gateway and switch models by changing one string. Start free, pay only when you top up.
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.llmgateway.io/v1", apiKey: process.env.LLM_GATEWAY_API_KEY,});
const res = await client.chat.completions.create({
model: "anthropic/claude-sonnet-5",
messages: [{ role: "user", content: "Hello" }],
});40+ providers on the network
View all providersEnterprise starts with a pilot on your own traffic. In week 1 we move your keys, routing rules and first production traffic with you. In week 2 SSO, audit logs and guardrails are set up and checked by your security team. At day 30 you decide. If we missed a milestone, you can walk away, with no long-term contract before that point.
Start your pilotBring your own provider keys at no platform fee. Credits are pay-as-you-go with a 5% platform fee on top-ups and no token markup; non-US cards may add a 1.5% international card fee. Optional full data retention costs $0.01 per 1M tokens in both modes. Enterprise pricing is custom, with volume discounts and a 99.9% uptime SLA on Enterprise Cloud.
See pricingYes. Run it with Docker or Kubernetes using our Helm chart, inside your own network. The core gateway is open source under AGPLv3 and free to self-host. Enterprise features such as SSO, audit logs and guardrails need an enterprise license.
Our public status page shows live status and 90-day uptime. Enterprise Cloud contracts include a 99.9% uptime SLA. Pay-as-you-go has no SLA, and self-hosted installs depend on your infrastructure.
Status pageLLM Gateway is open source: you can self-host the gateway under AGPLv3, while enterprise features need a license. Your own provider keys carry no platform fee at any volume, while OpenRouter charges 5% on bring-your-own-key usage above $25k of list-price inference a month ($200k on its enterprise plan). Stripe announced in August 2026 that it is acquiring OpenRouter.
Full comparison250+ models from 40+ providers, and most major releases go live within 48 hours of launch.
Browse all modelsEnterprise
Security questionnaire, DPA, SAML SSO setup and hands-on migration of your keys and routing rules. We handle onboarding with you.
Traffic live
SSO, audit logs and guardrails on
You decide. No long-term contract before then