AI Bridge for OpenAI-Compatible APIs

By linger0er

0+ active installationsVersion 0.1.6

Connect WordPress AI to custom OpenAI-compatible APIs with Responses, Chat Completions, a default model, and up to 10 models.

Description

AI Bridge for OpenAI-Compatible APIs registers an AI provider for the WordPress AI Client. It lets site administrators configure a shared API base URL, an API key through WordPress Connectors, and multiple text generation models for a service that exposes OpenAI-compatible endpoints.

This plugin is not affiliated with, endorsed by, sponsored by, or officially connected to OpenAI.

Each model can use one of two endpoint types:

  • Responses-compatible endpoint: /v1/responses
  • Chat Completions-compatible endpoint: /v1/chat/completions

For Responses-compatible models, the plugin adapts WordPress AI Client chat-style prompt payloads into the Responses API input format, then adapts Responses API output back into the chat-style response shape expected by the WordPress AI Client.

External Services

This plugin sends text generation requests to the API endpoint configured by the site administrator. The plugin does not contact an AI service until an administrator enters an API base URL and configures credentials.

Data sent to the configured service can include prompt text, system instructions, model configuration, and any other request fields provided by the WordPress AI Client for text generation.

If the configured endpoint is OpenAI’s API, usage is subject to OpenAI’s terms and privacy policy:

  • Terms: https://openai.com/policies/terms-of-use
  • Privacy policy: https://openai.com/policies/privacy-policy

If another OpenAI-compatible gateway is configured, usage is subject to that service provider’s terms and privacy policy.

Installation

  1. Install the plugin ZIP through the WordPress Plugins screen.
  2. Activate the plugin through the Plugins screen in WordPress.
  3. Open Settings > Connectors and set the API key for AI Bridge for OpenAI-Compatible APIs.
  4. Open Settings > AI Bridge for OpenAI-Compatible APIs.
  5. Configure the API base URL and enabled model list.
  6. Use the + Add model button to add more models (up to 10), select an enabled default model, and save.

FAQ

Does this plugin require an API key?

It requires an API key for the OpenAI-compatible service configured by the site administrator.

Where is the API key stored?

The API key is stored through the WordPress Connectors settings for this provider.

Can I use a local gateway?

Yes. Configure the base URL as a full HTTP URL such as http://localhost:1234/v1.

Can I choose a default model?

Yes. Select Default beside an enabled model on this plugin’s settings screen and save. The default is preferred by automatic text generation in the WordPress AI plugin and listed first within this provider. Explicit model or provider selections in individual features take precedence. Removing or disabling the default selects the first remaining enabled model.

How many models can I configure?

Up to 10. New installations show one model row; click + Add model for more rows. Existing model settings are retained. Model IDs must be unique.

Changelog

0.1.6

  • Restore English source descriptions for the official WordPress.org translation workflow.
  • Add locale-based Simplified Chinese translations for plugin metadata and settings.

0.1.5

  • Add Chinese plugin descriptions, feature explanations, and setup instructions.
  • Keep repository documentation and the license file at the repository root.

0.1.4

  • Rename the plugin to AI Bridge for OpenAI-Compatible APIs while preserving existing settings and update compatibility.
  • Add a default model for automatic text generation in the WordPress AI plugin.
  • Allow up to 10 models with add and remove controls, starting with one row.
  • Make model configuration changes available immediately without an API discovery cache.
  • Correct Responses assistant history and report interrupted streams, failed requests, and output token limits.
  • Apply the environment base URL consistently and respect explicitly configured request timeouts.

0.1.3

  • Add an optional per-model streaming setting for Responses API requests.
  • Aggregate streaming Responses API events into a standard text-generation result.

0.1.2

  • Clarify the public installation instructions.
  • Remove developer-only troubleshooting and local development details from the FAQ.

0.1.1

  • Remove the unsupported temperature parameter from Responses API requests.
  • Clarify that the plugin is licensed under GPLv2.

0.1.0

  • Initial release.