Configure OpenCat, ChatBox, and LobeChat with WaveSpeed LLM: 290+ Models in One App

The Three Fields Every Chat App Asks You

Every modern desktop/mobile LLM chat app — OpenCat, ChatBox, LobeChat, ChatGPT Next Web, Cherry Studio, and dozens more — asks you to fill in the same three fields when you add a custom provider:

1. Endpoint (sometimes called “API Host”, “Base URL”, or “Server URL”)
2. API Key (sometimes “Secret Key”, “SK”, or “Token”)
3. Chat Protocol (sometimes “API Type” or “Compatible Mode”)

For most people, those three blanks are where configuration stalls. This guide tells you exactly what to paste, for three of the most popular chat apps — OpenCat, ChatBox, and LobeChat — using WaveSpeed LLM as the backend.

The Universal Answers

No matter which chat app you’re configuring, for WaveSpeed LLM the answers are always:

Field	Value
Endpoint / Base URL	https://llm.wavespeed.ai/v1
API Key	Your WaveSpeed API key (from the dashboard)
Chat Protocol	OpenAI (Chat Completions)
Model ID	vendor/model — e.g. anthropic/claude-opus-4.6

With those four values you can connect any OpenAI-compatible client. Below we walk through three specific apps.

Prerequisite: Get Your WaveSpeed API Key

1. Go to wavespeed.ai and sign in.
2. Open Dashboard → API Keys.
3. Create a new key, copy it immediately, store as a secret.

Keep this tab open — you’ll paste the key in each app below.

OpenCat

OpenCat is a popular native chat client for macOS and iOS.

1. Open OpenCat → Settings → API Providers → Add Provider (or edit an existing OpenAI provider).
2. Fill in:
   • Name: WaveSpeed LLM (any label you like)
   • API Host: https://llm.wavespeed.ai/v1
   • API Key: YOUR_WAVESPEED_API_KEY
   • Protocol / Provider Type: OpenAI
3. Save. OpenCat will list available models — pick anthropic/claude-opus-4.6 or any other model from the WaveSpeed catalog.
4. If OpenCat doesn’t auto-list models, add one manually with ID anthropic/claude-opus-4.6.

Tip: OpenCat lets you create multiple assistants with different models. Point each one at a different WaveSpeed model ID to compare Claude, GPT, Gemini side by side.

ChatBox

ChatBox is a cross-platform (Windows, macOS, Linux, mobile) chat app.

1. Open ChatBox → Settings → Model.
2. Set:
   • Model Provider: OpenAI API (or “Custom OpenAI”)
   • API Key: YOUR_WAVESPEED_API_KEY
   • API Host: https://llm.wavespeed.ai/v1
   • API Path: /chat/completions (default, leave as is)
   • Model: anthropic/claude-opus-4.6 (type it in if not in the list)
3. Save and start a new chat.

ChatBox caches its model list per-provider. If models don’t show up, type the model ID manually — ChatBox will use it.

LobeChat

LobeChat is a fully-featured open-source chat UI, self-hostable or via lobehub.com.

1. Open LobeChat → Settings → Language Model → OpenAI.
2. Toggle Enable.
3. Fill in:
   • API Key: YOUR_WAVESPEED_API_KEY
   • API Proxy Address (or Base URL): https://llm.wavespeed.ai/v1
4. Under Model List, add a custom model:
   • Model ID: anthropic/claude-opus-4.6
   • Display name: anything you want
5. Save and pick the new model in a conversation.

LobeChat’s OpenAI provider is the right one to pick even when you’re calling Claude or Gemini through WaveSpeed — the client speaks OpenAI Chat Completions and WaveSpeed translates to upstream models transparently.

Why “OpenAI” is Always the Right Protocol Choice

A common point of confusion: if I’m calling Claude through WaveSpeed, shouldn’t I pick “Anthropic” as the protocol?

No. Pick OpenAI.

WaveSpeed LLM exposes every model — Claude, GPT, Gemini, Llama, Qwen, DeepSeek — through the OpenAI Chat Completions protocol. The anthropic/ prefix in the model ID just tells WaveSpeed which upstream model to route to. Your chat app only needs to speak OpenAI.

This is the whole point of a unified OpenAI-compatible gateway: your tool doesn’t need to know the difference.

Popular Model IDs to Try

Model ID	Notes
anthropic/claude-opus-4.6	1M context Claude
anthropic/claude-sonnet-4.6	Faster, cheaper Claude
openai/gpt-5.2-pro	Top OpenAI reasoning
openai/gpt-5.2-chat	Balanced GPT
google/gemini-3-flash-preview	Fast multimodal
deepseek/deepseek-v4	Cheapest strong coder
meta-llama/llama-4-70b	Open weights, steady

Browse the full catalog at wavespeed.ai/llm.

Troubleshooting

“Invalid API Key” / 401 Unauthorized You pasted the wrong key. Copy the key from the WaveSpeed dashboard exactly — no leading/trailing spaces.

“Model not found” / 404 The model ID must use the full vendor/model form. If your client auto-completes to a stripped form like claude-opus-4.6, edit it manually.

App tries to connect to api.openai.com You forgot to change the Base URL / API Host. Re-open settings and confirm it’s https://llm.wavespeed.ai/v1.

SSL / certificate errors Update your chat app to the latest version. All modern versions support standard TLS connections to llm.wavespeed.ai.

One Key, Many Apps

Add the same WaveSpeed API key to every chat client you use. One account, one bill, 290+ models everywhere.

Get your WaveSpeed API key and connect your favorite chat app in minutes.