Bestony AI Provider for Xiaomi MiMo

Description

Adds Xiaomi MiMo (小米 MiMo) as a provider for the WordPress AI Client.

• The model list is fetched live from GET /v1/models, so newly released models appear on their own.
• Text generation and chat history with the MiMo chat models, including tool calling, structured
  output (JSON mode) and stop sequences.
• Vision (multimodal understanding): the mimo-v2.6-* family and mimo-v2.5 accept images, so
  features like Alt Text Generation work.
• Reasoning output is surfaced as thought parts rather than mixed into the answer text.
• A Settings → Bestony AI Provider for Xiaomi MiMo page with three options: the API host (standard API or
  one of the Token Plan clusters), the thinking mode, and the structured-output mode.

Settings

Settings → Bestony AI Provider for Xiaomi MiMo has three options. Each is overridden by its matching
environment variable when one is set.

API host — which Xiaomi MiMo host the provider talks to:

• Xiaomi MiMo (standard API) — https://api.xiaomimimo.com/v1 (default)
• Token Plan — China — https://token-plan-cn.xiaomimimo.com/v1
• Token Plan — Singapore — https://token-plan-sgp.xiaomimimo.com/v1
• Token Plan — Europe — https://token-plan-ams.xiaomimimo.com/v1

Pick the standard API for a pay-as-you-go key, or the Token Plan cluster your subscription key was
provisioned for. The chosen host is used for the model list and for every generation request. It is
stored per site, so each site on a multisite network configures its own. Only these four hosts can be
selected: a value that is not one of them (written to the database directly, for example) falls back
to the default.

Thinking mode — Use the model default (default), Always enable thinking, or Disable thinking.
MiMo enables thinking by default; while it is on, the model forces its own temperature/top_p and
the answer arrives as reasoning_content plus content. Disable it for cheaper, faster calls that
honour those parameters.

Structured output — JSON mode (json_object) (default) or Send no response_format. MiMo
implements OpenAI JSON mode only, not schema-constrained output; a feature that asks for a JSON
schema is downgraded to json_object.

There is also a Setup Token Plan link in the plugin’s row on the Plugins screen that jumps
straight to this page.

The admin screens are translatable; a translation template (languages/bestony-ai-provider.pot) ships with the plugin.

Configuration

The API key is read, in order of precedence, from:

1. The XIAOMIMIMO_API_KEY environment variable
2. The XIAOMIMIMO_API_KEY PHP constant
3. The connectors_ai_xiaomimimo_api_key option (Settings → Connectors)

Each settings-page option is overridden by its environment variable, and the remaining behaviour is
controlled by environment variables or PHP constants too:

• XIAOMIMIMO_DEFAULT_MODEL — chat model to prefer in pickers and feature filters.
  Default: mimo-v2.6-pro.
• XIAOMIMIMO_BASE_URL — API base URL. Overrides the host chosen on Settings → AI Provider for
  Xiaomi MiMo.
• XIAOMIMIMO_THINKING — default, enabled or disabled. Overrides the Thinking mode setting.
• XIAOMIMIMO_STRUCTURED_OUTPUT — json_object or none. Overrides the Structured output setting.
• XIAOMIMIMO_MODEL_INPUT_MODALITIES — comma-separated list of modalities your models accept. Include
  image (e.g. text,image) to declare vision for every chat model, for deployments ahead of
  the built-in list.
• XIAOMIMIMO_REQUEST_TIMEOUT — request timeout in seconds. Default: 120.
• XIAOMIMIMO_CONNECT_TIMEOUT — connection timeout in seconds. Default: 10.

Because the model list is cached by the AI Client, changing a setting or an environment variable does
not invalidate that cache. If a new value does not take effect immediately, clear the cache — in
WP-CLI:

wp cache flush
wp transient delete --all

To influence which models the AI plugin picks, use the standard filters in your own plugin or theme —
this plugin already puts its preferred model first:

add_filter( 'wpai_preferred_text_models', function ( $models ) {
    array_unshift( $models, array( 'xiaomimimo', 'mimo-v2.6-flash' ) );
    return $models;
} );<h3>External services</h3>

This plugin connects to the Xiaomi MiMo API, an external service operated by Xiaomi (小米). It is
required so the WordPress AI Client can send requests to MiMo models from your site. MiMo is a paid
service: requests are billed to your Xiaomi MiMo account, and an account with API access is required.

The plugin contacts the following endpoints under the selected base URL — https://api.xiaomimimo.com/v1
by default, or whichever host is chosen on Settings → Bestony AI Provider for Xiaomi MiMo:

• GET /models — called when the AI Client refreshes its list of available models, and when it checks
  whether your credentials work. No user content is sent; only your API key, so MiMo can return the
  models available to your account.
• POST /chat/completions — called whenever any plugin or theme on your site uses the WordPress AI
  Client to generate text or to analyze an image. The request carries your API key and the prompt:
  messages, system instruction, tool definitions, and any other parameters the calling code supplied
  (conversation history, a JSON schema for structured output, or images attached to the prompt).

No request is made until something on your site asks the AI Client for a generation, or the AI Client
refreshes its model list. Your API key is stored on your own site and is only ever sent to the host
configured above.

This service is provided by Xiaomi:

• Platform and documentation: https://platform.xiaomimimo.com/
• Documentation hub: https://mimo.mi.com/docs/
• Terms of Service: https://mimo.mi.com/docs/quick-start/terms/user-agreement
• Privacy Policy: https://mimo.mi.com/docs/quick-start/terms/privacy-policy

Technical details