Skip to main content
Most AI apps let you point them at a different inference endpoint. Opper’s gateway speaks both the OpenAI and Anthropic wire formats, so once an app can take a base URL and an API key, it runs on any of the models in the catalog — with one key, EU or US hosting, and every call visible in traces.
Seven agents need no configuration at all. The Opper CLI writes their config for you:

​
One base URL

Every integration on this page points at the same endpoint:
It serves both compatibility surfaces from that one URL — OpenAI Chat Completions and Responses, and the Anthropic Messages API. Apps that append their own path (/chat/completions, /v1/messages) work unchanged; see Drop-in SDKs for the full surface. Model IDs use the provider/model convention — anthropic/claude-sonnet-4-6, openai/gpt-5.5, gemini/gemini-3.5-flash. Apps that support model discovery call GET /v3/compat/models and populate their own picker, filtered by your model access rules, so a model your policy denies is never offered.

​
Launch from the CLI

For these agents, one command provisions the config and starts the app. Nothing to edit by hand: Pin a model with --model, and remove any integration later with opper agents uninstall <agent> — it unwires Opper without uninstalling the app.

​
Coding agents

Claude Code

Anthropic’s terminal coding agent, on any model

Codex

OpenAI’s open-source coding CLI

OpenCode

SST’s open-source terminal coding agent

pi

Terminal agent with a native Opper provider

DeepSeek Harness

DeepSeek’s plugin-first agent harness

GitHub Copilot CLI

GitHub’s terminal agent, no Copilot plan needed

Aider

AI pair programming in your terminal

Goose

Open agent from the Agentic AI Foundation

Droid

Factory’s agent for terminal and IDEs

Voidleap Code

Desktop agentic IDE with Opper built in

​
Editors and IDEs

GitHub Copilot

Copilot Chat and agent mode, via the Opper VS Code extension

Cursor

The AI code editor

Cline

Open-source AI coding agent for VS Code

Kilo Code

Open-source AI agent for VS Code

Zed

The high-performance code editor

JetBrains AI Assistant

IntelliJ, PyCharm, and every JetBrains IDE

Continue.dev

Open-source assistant for VS Code and JetBrains

​
Personal agents

Claude Cowork

Anthropic’s agent for knowledge work

Claude Desktop

Anthropic’s macOS and Windows desktop app

Copilot for Obsidian

AI copilot inside your Obsidian vault

OpenClaw

Personal agent on WhatsApp, Telegram, and Discord

Hermes

Nous Research’s open terminal agent

​
Chat

LibreChat

The open-source AI chat platform

Open WebUI

Self-hosted AI workspace

​
Automation

n8n

Workflow automation with a native Opper node

​
Frameworks and gateways

Not apps — libraries and proxies you build on or run in front of Opper.

Vercel AI SDK

TypeScript toolkit with an official Opper provider

Mastra

TypeScript agent framework with a first-class opper/ provider

LiteLLM

Use Opper as the upstream router for a self-hosted gateway
Opper is listed on models.dev, the open catalog many tools read their model list from. If an app pulls its providers from there, Opper may already be in its picker — with model IDs, context windows, and pricing filled in — before you configure anything.

​
Why route an app through Opper

​
Your app isn’t listed

Most of these integrations are the same three fields. If your app has a setting for a custom OpenAI-compatible provider, it already works:
1

Set the base URL

If the app expects a host and a path separately, the host is https://api.opper.ai and the path is v3/compat/chat/completions.
2

Paste an Opper API key

Create one at platform.opper.ai. Keys are scoped to a project, so use a separate one per app if you want usage split out.
3

Pick a model

Use a full model ID such as anthropic/claude-sonnet-4-6. Apps that fetch GET /v3/compat/models will list the catalog for you.
Tag the traffic with X-Opper-Tags to break usage down per app — see Tags and usage attribution.

​
Next steps

Browse the model catalog

Model IDs, hosting regions, and retention metadata.

Opper CLI

Launch agents, inspect traces, and manage models from the terminal.