Integrations
LLMRPM speaks two wire formats: an OpenAI-compatible API and an Anthropic-compatible API. Any client that lets you set a custom base URL and a bearer key can talk to it. Pick your tool below for a step-by-step recipe.
Supported tools
Each guide covers configuration, a working example, recommended models, and troubleshooting for one tool. Every recipe uses your key from /dashboard/api-keys and one of the two LLMRPM base URLs.
OpenAI SDK
Point the official openai package at LLMRPM in Node or Python.OpenAI-compatiblecURL / HTTP
Talk to LLMRPM directly over HTTPS from any language or shell.EitherCursor
Use LLMRPM as a custom OpenAI-compatible model in Cursor.OpenAI-compatibleClaude Code
Route Claude Code at LLMRPM's Anthropic-compatible endpoint.Anthropic-compatibleWindsurf
Add LLMRPM as an OpenAI-compatible custom provider in Windsurf.OpenAI-compatibleCline
Configure Cline against LLMRPM in Anthropic- or OpenAI-compatible mode.Anthropic-compatibleRoo Code
Set up Roo Code with LLMRPM's OpenAI-compatible endpoint.OpenAI-compatibleKilo Code
Point Kilo Code's OpenAI-compatible provider at LLMRPM.OpenAI-compatibleContinue
Add LLMRPM models to the Continue VS Code / JetBrains extension.OpenAI-compatibleOpenCode
Override OpenCode's OpenAI provider to reach LLMRPM.OpenAI-compatibleAmp
Bridge Amp to LLMRPM through a local CLIProxyAPI proxy.Anthropic-compatibleAider
Run the Aider CLI against LLMRPM's OpenAI-compatible endpoint.OpenAI-compatibleZed
Expose LLMRPM models in Zed's assistant model picker.OpenAI-compatibleJetBrains
Configure a JetBrains AI plugin to use LLMRPM as an OpenAI provider.OpenAI-compatibleHow it works
Every integration configures one of two base URLs with your LLMRPM key. If a tool lets you override the base URL of the OpenAI or Anthropic client it already uses, it works with LLMRPM.
| API shape | Base URL | Used by |
|---|---|---|
| OpenAI-compatible | https://llmrpm.com/api/v1 | OpenAI SDK, Cursor, Windsurf, Continue, Zed, most tools |
| Anthropic-compatible | https://llmrpm.com/api | Anthropic SDK, Claude Code, Cline (Anthropic mode) |
Bearer auth, both formats
LLMRPM authenticates every request with
Authorization: Bearer llmrpm_.... For Anthropic SDKs set the authToken field (not apiKey) so the client sends a bearer token instead of an x-api-key header.Supported surface
LLMRPM exposes chat completions, Anthropic messages, a model list, and usage endpoints. Features that depend on other endpoints — embeddings, image or audio generation, or provider-specific extensions — are not part of this surface, so configure integrations for chat only.
Environment variables
The examples reference a small, consistent set of variables. Set only the ones your integration uses:
| Variable | Value |
|---|---|
| LLMRPM_API_KEY | Your key from /dashboard/api-keys, e.g. llmrpm_.... |
| LLMRPM_BASE_URL | https://llmrpm.com/api/v1 — for OpenAI-compatible clients and raw HTTP. |
| LLMRPM_ANTHROPIC_BASE_URL | https://llmrpm.com/api — for Anthropic SDKs only; see Messages. |
.env
LLMRPM_API_KEY=llmrpm_your_key_here
LLMRPM_BASE_URL=https://llmrpm.com/api/v1
LLMRPM_ANTHROPIC_BASE_URL=https://llmrpm.com/apiOne key per environment
Issue separate keys for local, staging, and production from
/dashboard/api-keys. Revoking a leaked key then affects only that environment, and per-key usage stays legible in your usage view.Next steps
- Browse the model list for public IDs to drop into any recipe.
- Read Chat completions and Messages for the full request and response shapes.
- Check Usage & limits to understand the 5-hour, 24-hour, and 7-day request windows.