DuoPort Connector for OpenCode

DuoPort Connector for OpenCode registers two WordPress AI Client providers:

• OpenCode Go for the subscription catalog.
• OpenCode Zen for the pay-as-you-go catalog, including reviewed free-model records.

Enter your opencode.ai key in the Go and Zen fields on Settings → Connectors. In most cases, the same key works for both catalogs, while each field remains independently testable. DuoPort keeps credentials in WordPress and keeps model policy in a reviewed registry.

This plugin is not affiliated with or endorsed by OpenCode (Anomaly Innovations, Inc.).

Why DuoPort?

DuoPort gives WordPress teams a small, auditable OpenCode bridge instead of another opaque AI dashboard. It works with the native WordPress AI Client, keeps Go and Zen catalog identities separate, and exposes only model routes whose endpoint and capability contracts have been reviewed. The optional Model Radar tracks public catalog changes and queues new or free-model candidates for verification; it never auto-promotes a model.

OpenCode Go and Zen

OpenCode Go

Use the Go subscription catalog for its reviewed text-generation models. Go availability is checked with a bounded probe and transient cache, so the plugin can report configured or unavailable states without repeatedly calling the catalog API.

OpenCode Zen

Use the Zen pay-as-you-go catalog for reviewed text-generation models, including records explicitly identified as free. Free availability is supplied by OpenCode and can change; a free-looking model name is only a verification candidate, not a promise of free access.

Current OpenCode models

Model lists change over time. By default, DuoPort presents the reviewed chat/completions records for each catalog and labels Zen free records. The credential-free Model Radar compares current public Go and Zen /models evidence with that registry and reports new, retired, unsupported, and verification-required changes in one aggregate issue.

Enable Show all models only when you want to inspect the complete live catalog. Unknown models, unknown endpoint families, unsupported capabilities, and unverified free-name candidates remain default-deny; discovering a model is not the same as making it routable.

Free OpenCode models

Zen records matched against OpenCode’s published pricing table are marked (Free) in the model picker when they are part of the reviewed registry. Free models still require an API key and may have limits, availability changes, or catalog retirement. DuoPort does not promise that every free-looking ID is free or supported.

WordPress AI Client integration

DuoPort uses the native WordPress AI Client provider registry. It does not create a separate site-wide model router, duplicate WordPress credentials, or read connector secrets for diagnostics. Model selection, metadata, and request transport remain inside the plugin’s reviewed boundaries.

Verified model support

Only reviewed chat/completions records are advertised by default. The current adapter does not claim Responses, Messages, image, tool, or web-search support unless the corresponding contract is explicitly present in the registry. Unknown or unresolved routes fail clearly instead of silently falling back.

Features

• Two providers (opencode-go / opencode-zen) auto-discovered by the Connectors screen.
• Reviewed chat/completions-routable models per catalog (Go: 11, Zen: 15) — present in the live OpenCode /models catalog and recorded as routable in the reviewed registry.
• Explicitly reviewed free models labeled (Free) in the model picker (Zen catalog).
• Credential-free Model Radar and catalog watch for fast-follow verification work.
• Availability probe with transient caching — validates key without calling /models.
• Show all models toggle on Settings → DuoPort Connector (off by default).

External Services

This plugin connects to the OpenCode API (https://opencode.ai) to list models, check availability, and generate text.

• Your API key is sent with every request.
• Prompts and messages you submit for generation are sent to OpenCode’s servers.
• Model list is fetched from https://opencode.ai/zen/go/v1/models and https://opencode.ai/zen/v1/models.
• Availability is validated by a chat/completions probe (max_tokens: 1).
• See https://opencode.ai/legal/terms-of-service for terms and https://opencode.ai/legal/privacy-policy for privacy policy.