logo
Sign In
preswest
@preswest
·
9 months ago
·
a year ago
function
OpenRouter Integration for OpenWebUI
Last Updated
9 months ago
Created
a year ago
Function
pipe
v0.4.1
Name
OpenRouter Integration for OpenWebUI
Downloads
4.1K+
Saves
19+
Description
Access the full suite of OpenRouter models directly within OpenWebUI with support for citations and reasoning tokens.

Version: v0.4.1 Released: 2025-05-25 Changes: Added "free only" model filter. Github: https://github.com/kevarch/OpenWebUI-Openrouter-API-Model-Function

OpenWebUI-Openrouter-API-Model-Function

This pipe integrates the OpenRouter.ai API with OpenWebUI, allowing users to access a wide variety of large language models available through OpenRouter directly within the OpenWebUI interface.

It acts as a bridge, fetching the list of available models from OpenRouter and forwarding chat requests from OpenWebUI to the selected OpenRouter model, handling both streaming and non-streaming responses.

Getting started with OpenRouter is straightforward: simply sign up on their website at OpenRouter.ai, add some credits, and generate an API key to use with this function.

Features

  • Access to Diverse Models: Leverages the OpenRouter API to provide access to numerous LLMs from various providers.
  • Model Discovery: Automatically fetches and lists available OpenRouter models within OpenWebUI.
    • New models will appear automatically as they are released on OpenRouter.
    • Optionally specify a list of providers to include or exclude, useful to avoid duplication with models already integrated in OpenWebUI or to focus on specific providers.
  • Optional Model Prefix: Allows adding a custom prefix (e.g., OR: ) to model names in the OpenWebUI list for easy identification. This field is optional and can be left empty for more flexible configuration.
  • Free Models Filter (FREE_ONLY): Enables showing only free models to facilitate testing and usage without consuming credits.
  • Streaming & Non-Streaming Support: Handles both response types seamlessly.
  • Citation Support:
    • Fetches citation data provided by compatible OpenRouter models.
    • Formats citations into a readable list appended at the end of the response.
    • Replaces citation markers (e.g., [1]) with clickable Markdown links (e.g., [[1]](citation_url)).
  • Reasoning Token Support:
    • Optionally requests reasoning/thought processes from models that support this feature (include_reasoning=true).
    • Wraps reasoning output in <think>...</think> tags for clear separation in the UI.
  • Configurable Timeout: Set a custom timeout for API requests to OpenRouter.
  • Error Handling: Provides informative error messages in the UI for common issues like invalid API keys, network timeouts, or API errors.
  • Optional Prompt Caching: Allows enabling OpenRouter's prompt caching feature (cache_control), potentially reducing costs for supported models on repeated requests with similar initial prompts.
  • Optional Model Providers (MODEL_PROVIDERS): This parameter accepts a comma-separated list of providers to include or exclude and is optional for more flexible setup.

Installation

This function can be installed manually or easily installed from the OpenWebUI Hub:
https://openwebui.com/f/preswest/openrouter_integration_for_openwebui

After installing:

  • Configure the function:
    • Log in to OpenWebUI as an admin user.
    • Navigate to Admin Panel -> Functions.
    • Click the Settings (gear icon ⚙️) button for the function.
    • Enter your OpenRouter API Key (required).
    • Adjust other settings such as Include Reasoning, Model Prefix, Request Timeout, FREE_ONLY, etc., as desired.
    • Click Save.

Configuration

The following settings can be configured via the OpenWebUI Pipe settings interface:

  • OPENROUTER_API_KEY (Required): Your personal API key from OpenRouter.ai.
  • INCLUDE_REASONING (Default: True): Whether to request reasoning tokens from models that support them. If enabled, reasoning appears within <think> tags.
  • MODEL_PREFIX (Optional, Default: ""): An optional text prefix added to the names of OpenRouter models displayed in OpenWebUI (e.g., setting it to OR: will show models like OR: xAI: Grok 3 Beta).
  • REQUEST_TIMEOUT (Default: 90): The maximum time (in seconds) to wait for a response from the OpenRouter API before timing out.
  • MODEL_PROVIDERS (Optional): Comma-separated list of model providers to include or exclude. Leave empty to include all providers.
  • INVERT_PROVIDER_LIST: If true, the 'Model Providers' list becomes an exclude list instead of an include list.
  • FREE_ONLY (Optional, Default: False): If true, filters to show only free models.
  • ENABLE_CACHE_CONTROL (Default: False): If true, enables OpenRouter's prompt caching, potentially reducing costs for supported models on repeated requests with similar initial prompts and repeated context.

Usage

  1. After installation and configuration, the models fetched from OpenRouter (potentially with your prefix) will appear in the model selection dropdown in OpenWebUI.
  2. If desired, enable the filter for free models with FREE_ONLY.
  3. Select an OpenRouter model.
  4. Chat as usual. The pipe will handle communication with the OpenRouter API.
  5. If the model provides citations or reasoning (and reasoning is enabled), they will be formatted and included in the response.

License

This project is licensed under the MIT License.

4