What is LLM Gateway?
One OpenAI-compatible endpoint for many model providers, with spend tracking, failover and caching.
Who is LLM Gateway best for?
LLM Gateway is best for product teams that use several LLM providers and want one API with cost tracking.
Switching models without rewrites
It accepts OpenAI-style requests, so moving to another provider means changing the base URL and the model name.
Watching LLM spend
Dashboards split requests, tokens and cost by model, provider and API key, and spend caps stop a key at its limit.
Staying up when a provider fails
When a provider errors, times out or rate-limits, the gateway retries the same model on another provider, up to twice.
Keeping traffic in-house
The AGPLv3 core runs inside your own network with Docker or Kubernetes.
How much does LLM Gateway cost?
LLM Gateway is free to self-host and free in its cloud with your own provider keys; paying through its credits adds a 5% fee to each provider's list price.
- Run the gateway with Docker or a Helm chart
- Your data stays on your servers
- Enterprise features need a license key
- Bring your own provider keys with no platform fee
- Routing, caching, failover and spend caps
- Metadata-only logs, 20 requests a minute on free models
- 200+ models at provider list prices
- No rate limit on paid models
- Non-US cards add a 1.5% fee
Enterprise is priced by sales after a 30-day trial. It adds volume discounts on credits, SSO, guardrails, audit logs, a 99.9% uptime SLA and 30 days of full request and response retention, which is billed at $0.01 per million tokens. Credit purchases with less than 10% used can be refunded within 14 days. Full pricing on llmgateway.io. We checked these prices on .
How do readers rate LLM Gateway?
No one has rated LLM Gateway yet. Be the first.
Sign in to rate. One rating per account.
Is LLM Gateway open source?
Yes. LLM Gateway's code is public under the AGPL 3.0 licence, so you can read it, change it and run it yourself.
On GitHub
1,736stars
- Last commit
- 12 Nov
- Repository age
- 1 years
- Latest release
- 9 Nov
- Licence
- AGPL 3.0
- Repository
- theopenco/llmgateway
How do I connect LLM Gateway to my AI agent?
LLM Gateway MCP is LLM Gateway's official MCP server. Add it to your agent once, then ask it to track spend and model usage in your LLM Gateway account.
Gives your agent read access to your LLM Gateway stats: how much you spent, which models and providers get the most calls and which coding apps send them.
$ claude mcp add --transport http llm-gateway https://api.llmgateway.io/mcp
Claude Code connects to LLM Gateway over the web.
Run /mcp and sign in to LLM Gateway in the browser.
Then ask your agent
- Runs
- On the web
- Sign-in
- OAuth
- Checked
- Answered 13 Nov
- Guide
- docs.llmgateway.io
What are the best LLM Gateway alternatives?
Three tools on Recaply do LLM Gateway's job.
Questions about LLM Gateway
What is LLM Gateway?
One OpenAI-compatible endpoint for many model providers, with spend tracking, failover and caching.
Who is LLM Gateway for?
LLM Gateway is best for product teams that use several LLM providers and want one API with cost tracking.
Is LLM Gateway free?
Yes. LLM Gateway has a free Self-hosted plan: run the gateway with Docker or a Helm chart.
How much does LLM Gateway cost?
LLM Gateway's paid plans are Credits at 5% per credit top-up.
Is LLM Gateway open source?
Yes. LLM Gateway's code is public under the AGPL 3.0 licence, so you can read it, change it and run it yourself.
Does LLM Gateway have an MCP server?
Yes. LLM Gateway MCP is LLM Gateway's official MCP server, so agents such as Claude Code, Cursor, Codex and GitHub Copilot can track spend and model usage in your LLM Gateway account. LLM Gateway runs it on the web and you sign in with your LLM Gateway account.
What are the best LLM Gateway alternatives?
The closest alternatives on Recaply are LiteLLM, Portkey and APIPark.
Where can I try LLM Gateway?
On its official website, llmgateway.io.
Can I self-host LLM Gateway?
Yes. LLM Gateway's core is open source under AGPLv3 and runs on your own servers with Docker or a Kubernetes Helm chart. Enterprise features such as SSO, guardrails and audit logs live in separate commercial code and need a license key.







