Skip to content
RunAPI Developer Docs
EN 中文
Developer Resources
Developer Resources

AnythingLLM

Install AnythingLLM, configure its Generic OpenAI provider for RunAPI, choose model limits, and verify a workspace chat.

View Markdown

AnythingLLM exposes a developer-focused OpenAI (generic) provider for compatible Chat Completions services. Before starting, create a standard RunAPI API key in the Authentication Guide and choose a model from the Model Catalog.

Install AnythingLLM

Install the desktop or self-hosted edition using the official AnythingLLM installation guide. Open AnythingLLM and finish its initial setup before changing the LLM provider.

Configure OpenAI (generic)

  1. Open Settings → LLM Preference and select OpenAI (generic).
  2. Enter https://runapi.ai/v1 as the Base URL. Do not append /chat/completions; AnythingLLM adds the operation path.
  3. Enter your RunAPI API key.
  4. Choose the exact RunAPI identifier under Selected Model. If the model list is empty, type the identifier from the Model Catalog.
  5. Set Model context window and Max Tokens to values supported by that model, then save the provider.

AnythingLLM uses Model context window to budget system instructions, history, and user input. It sends Max Tokens as max_tokens, so these values must follow the selected model rather than a universal example.

Verify the connection

Create or open a workspace, start a chat, and send Reply with the word connected. A text reply confirms that the workspace is using the saved RunAPI provider and Selected Model.

Troubleshoot

  • No models are listed: verify the Base URL is exactly https://runapi.ai/v1, then enter a current model identifier manually.
  • The request is rejected for length: reduce workspace history or correct Model context window and Max Tokens using the selected model’s current limits.
  • Authentication fails: update the API key in OpenAI (generic) with a valid standard key from the Authentication Guide.
  • Another model answers the workspace: reopen LLM Preference, confirm OpenAI (generic) is active, and save the intended Selected Model again.

For the shared protocol behavior, use the LLM API Quickstart. For request and response fields, use the Chat Completions API Reference.