LLMBase|Docs

Integrations

Geeps App

Connect Geeps App to LLMBase with an OpenAI-compatible custom provider.

Updated

Geeps App is a native BYOK chat client for iPhone, iPad, and Mac. Because Geeps supports custom OpenAI-compatible providers, you can use it with the LLMBase Inference API by adding your LLMBase API key, host, and model ID.

This guide configures Geeps for direct LLMBase inference. Use an inference key with the llmbase_... prefix, not a llmbase_chat_... agent key.

Prerequisites

  • Geeps App installed on macOS or iOS
  • An LLMBase inference API key
  • A model ID from the LLMBase model catalog

Configuration values

Geeps fieldValue
Provider nameLLMBase
API keyYour llmbase_... inference key
Base URLapi.llmbase.ai
Full custom endpointLeave empty
Default modelAny model ID returned by the LLMBase model catalog

Geeps builds the chat-completions endpoint from the base URL, so entering api.llmbase.ai produces:

https://api.llmbase.ai/v1/chat/completions

Add LLMBase as a provider

  1. Open Geeps and go to Settings.
  2. Select API Settings.
  3. Add or open a custom OpenAI-compatible provider.
  4. Enable the provider.
  5. Enter your LLMBase inference API key.
  6. Set the provider name to LLMBase.
  7. Set Base URL to api.llmbase.ai.
  8. Leave Full custom endpoint empty unless you intentionally need to override the generated endpoint.

Geeps API settings showing LLMBase provider name, API key, and base URL

Load and choose a model

Scroll to the Model section after the host is configured.

  1. Click Reload next to Get LLMBase models.
  2. Choose a default model from the model picker.
  3. If you prefer to enter it manually, paste a model ID exactly as shown in the model catalog.
  4. Click Submit under Test API Key to verify that Geeps can reach LLMBase.

Geeps model settings showing the LLMBase model reload button and test API key action

Start chatting

Return to the chat view and select the LLMBase provider or default model. Geeps will send OpenAI-compatible chat requests to:

POST https://api.llmbase.ai/v1/chat/completions

Requests spend your LLMBase prepaid inference balance. If a model does not support a feature that Geeps enables, such as vision or tools, choose a model with matching capabilities in the model catalog.

Troubleshooting

ProblemFix
Authentication failsConfirm the key starts with llmbase_ and has not been copied with spaces.
Endpoint preview looks wrongUse api.llmbase.ai as the base URL and leave the full custom endpoint empty.
Model reload failsEnter a model ID manually from the catalog, then test the API key.
A request is rejectedCheck that the selected model supports the requested feature, such as vision, tools, or structured output.

For the underlying API behavior, see OpenAI compatibility and Chat completions.