APIMODELSAPIMODELS

How to connect LobeHub to apimodels.app

TL;DR In LobeHub open Settings, Provider, Add Custom Provider. Give it a Provider ID and name, set Request Format to OpenAI, Proxy URL to https://api.apimodels.app/v1, and paste your apimodels.app key. After creating it, click Fetch models, enable the language models you want, and run the Connectivity Check.

LobeHub, formerly LobeChat, is an open-source chat client you can use hosted or run yourself. It lets you create your own provider from an address and a key, which is how you use apimodels.app in it.

The labels below were read from the LobeHub source and language files on 17 November 2026 (version 2.3.3). Later versions may rename things.

Create the provider

Go to Settings, then Provider, and click Add Custom Provider. The Create Custom AI Provider dialog opens.

FieldWhat to enter
Provider IDapimodels (lowercase letters, digits, - and _ only; it cannot be changed later)
Provider NameAny name, for example apimodels
Request FormatOpenAI
Proxy URLhttps://api.apimodels.app/v1
API KeyYour apimodels.app key

Which request format to choose

Choose OpenAI. It covers every language model on your key, Claude and Gemini included, because apimodels.app serves them all in the OpenAI chat format. The address must start with https:// and, for this format, should include /v1 as the placeholder in the dialog shows.

If you want Claude models to go through the Anthropic message format, create a second provider with Request Format set to Anthropic and Proxy URL https://api.apimodels.app. LobeHub removes a trailing /v1 for this format, so either form of the address works.

Fetch and enable models

Open the new provider, go to Model List, and click Fetch models. You can also click Add Model and type a Model ID such as claude-opus-5-5.

A model must be enabled before it appears in the chat picker. The fetched list includes image, video and audio models; enable only the language models.

For a model you add by hand, fill in Maximum Context Window and tick what the model supports: Support for Tool Calling, Support Vision, Support Deep Thinking. These switches only turn the feature on in the interface. Whether it works depends on the model.

Model IDMaximum Context WindowTool CallingVision
claude-opus-5-51,000,000YesYes
qwen4-max1,000,000YesYes

Run the Connectivity Check

On the provider page, find Connectivity Check, pick a model, and click Check. LobeHub sends one real chat request containing the word "hello" to that model, which uses a tiny amount of balance. "Check passed" means the address, key and model all work.

Limits and things that go wrong

Leave "Use Responses API Specification" off. LobeHub describes it as supported by OpenAI models only. The chat format works for every model here.

Leave "Use Client Request Mode" off unless you know you need it. It sends requests straight from your browser instead of through the LobeHub server.

The Provider ID is permanent. Pick it carefully; to change it you have to create a new provider.

Thinking models start slowly. qwen4-max finishes thinking before it writes, so the first words can take tens of seconds.

Image and video models are not for the chat window. Call them through the API. See the model pages for the request format.

Sources

Labels and behaviour were read from the LobeHub repository (canary branch, version 2.3.3) on 17 November 2026: the create-provider dialog, the request-format options, the Anthropic address handling, the connectivity checker and the language files. We did not test every LobeHub version.

Início rápido

cURL

# Check the address and key outside LobeHub
curl https://api.apimodels.app/v1/chat/completions \
  -H "Authorization: Bearer $APIMODELS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-opus-5-5",
    "messages": [{"role": "user", "content": "hello"}]
  }'

Perguntas frequentes

What Proxy URL do I enter in LobeHub?

With Request Format set to OpenAI, enter https://api.apimodels.app/v1. With Anthropic, enter https://api.apimodels.app.

Why can I not see my models in the chat picker?

Fetched and added models are off until you enable them. Enable the ones you want in the provider's Model List.

Is LobeHub the same as LobeChat?

Yes. The project was renamed and its repository is now lobehub/lobehub. The steps here apply to version 2.3.3.

Does the Connectivity Check cost anything?

Yes, a very small amount. It sends one real request with the word "hello" to the model you select.

Relacionado

Presente em
APIMODELS on Product HuntAPIMODELS featured on ToolifyAPIMODELS featured on There's An AI For That