README

AI Provider for MiniMax

MiniMax's frontier text models, inside WordPress. An independent MiniMax provider for the WordPress AI Client — brings the MiniMax M3 flagship and the full M2 series to every AI-enabled plugin on your site.

Not affiliated with, endorsed by, or sponsored by MiniMax.

What it does

WordPress 7.0 ships a built-in AI Client SDK: plugins ask it to "generate text from this prompt" without caring which AI service answers. This plugin registers MiniMax as one of those services, so its high-performance, cost-effective text models — the M2 and M3 series — become available to every AI-enabled plugin on your site.

Activate it, paste your MiniMax key, and you are generating content through frontier-quality models with strong multilingual support — no code required.

Features

Frontier models, one key

MiniMax M3 flagship (long-context) plus the full M2 series — M2.7, M2.5, M2.1, and M2, each with an optional Highspeed variant tuned for lower latency (8 models)
Live model discovery from the MiniMax API (cached for one hour), with a built-in 8-model fallback so generation keeps working if the API is briefly unreachable
MiniMax-M3 preselected as the default model out of the box

Native WordPress integration

Registers as a first-class provider for the WordPress 7.0+ AI Client — any AI-enabled plugin uses it with zero extra wiring
Enter your key on Settings → MiniMax, the core Settings → Connectors screen, or via the MINIMAX_API_KEY environment variable
Connection-status indicator on the settings page and the provider logo on the Connectors screen
Standard generation controls: temperature, max tokens, top P, and presence/frequency penalties

Built to stay out of the way

OpenAI-compatible under the hood — text generation, system instructions, stop sequences, and function declarations
Ships as both a WordPress plugin and a Composer package for standalone PHP use
No SDK bloat — WordPress core provides the AI Client at runtime, and the plugin excludes its own copy to stay collision-safe

Requirements

PHP 7.4+
WordPress 7.0+ (the AI Client SDK ships in core)
- On older WordPress, install the WordPress AI Client plugin separately

Installation

Method	How
WordPress.org	Search "AI Provider for MiniMax" under Plugins → Add New, then install and activate
Upload	Download the zip and upload it via Plugins → Add New → Upload Plugin
Composer	`composer require mralaminahamed/ai-provider-for-minimax`

Quick start

Activate the plugin.
Open Settings → MiniMax (or the core Settings → Connectors screen) and paste your API key. Get one at platform.minimax.io/user-center/basic-information/interface-key.
Any AI-enabled plugin can now generate text through the minimax provider:

use WordPress\AiClient\AiClient;

echo AiClient::prompt('Explain quantum computing')
    ->usingProvider('minimax')
    ->generateTextResult()
    ->toText();

See docs/USAGE.md for the standalone-Composer setup and per-request options.

Configuration

Configure defaults on Settings → MiniMax (option key minimax_settings):

Setting	Range	Default
Default Model	any available model	`MiniMax-M3`
Temperature	0.0–2.0	0.7
Max Tokens	1–200,000	4096
Top P	0.0–1.0	1.0
Presence Penalty	-2.0–2.0	0.0
Frequency Penalty	-2.0–2.0	0.0

The MINIMAX_API_KEY environment variable takes priority over the stored key. Full resolution order is in docs/ARCHITECTURE.md.

Supported models

When a key is configured, the live model list comes straight from the MiniMax API, so you always see the latest models. If the API is unavailable, an 8-model fallback keeps everything working — the MiniMax-M3 flagship plus the M2 series (M2.7, M2.5, M2.1, and M2), each with an optional Highspeed variant.

The complete, ID-by-ID catalogue and guidance on which model to pick are in docs/MODELS.md.

How it works

WordPress 7.0's AI Client is a provider-agnostic layer. Install one or more provider plugins — like this one — and any plugin built on the AI Client can generate content through whichever provider is active, switchable from the Connectors screen with no per-plugin reconfiguration. This plugin handles the MiniMax side: credential resolution, live model discovery, and the OpenAI-compatible API calls. Details in docs/ARCHITECTURE.md.

Documentation

Doc	What's in it
Usage	WordPress + standalone Composer examples, API key resolution, generation options, credential filters
Models	How model discovery and caching work, and the full 8-model fallback catalogue
Architecture	Provider identity, file layout, request flow, runtime SDK dependency
Development	Build/test/lint/analysis scripts, the AI Client `replace` guard and dev-only stub install, CI and release

Release history: CHANGELOG.md.

Development

composer install
composer test        # PHPUnit
composer phpcs       # WordPress Coding Standards
composer phpstan     # static analysis
composer release     # build the distributable zip

Full workflow — including the AI Client replace guard and the dev-only stub install that keeps PHPStan and PHPUnit green — is in docs/DEVELOPMENT.md.

Contributing

Issues and pull requests are welcome:

Open a new issue to report a bug or request a feature
Submit a pull request — see docs/DEVELOPMENT.md for the local build and test workflow

Support

If this plugin is useful, star the repo and leave a review on WordPress.org. Questions and bug reports go in the issue tracker.

mralaminahamed / ai-provider-for-minimax

Maintainers

Package info

Fund package maintenance!

Statistics

Security