7. Model Providers

The Models page ships 41 provider presets (official APIs, aggregators, Chinese and global cloud platforms) and also supports adding custom platforms.

7.1 Browse & filter

• The search box at the top matches platform names and model names
• Filter by plan (Coding Plan / Token Plan / Go Plan / Agent subscription / API only) or by status (authenticated / needs setup / needs verification / needs review / in use)
• Every platform card carries an auth status badge — hover for details (OAuth vs API key auth, last review time, …)

7.2 Configure a platform

Open a platform's edit form (custom platforms start from + Add platform):

• Endpoints: configure several; each row is a protocol type (anthropic / openai / responses) + Base URL. Custom platforms can switch an openai endpoint between chat / responses under "Advanced protocol settings"
• Auth: an API key (picked from the vault, can create one on the spot) or no auth (only for local services, public endpoints, or trusted internal gateways). Agent-native platforms like Claude subscriptions take no endpoint — they use OAuth login instead
• Model list: auto-filled after a successful connection test; you can also add/remove rows manually (each row is the model parameter used in requests)
• The form header links to the platform's official API docs and console

7.3 Connection test & model refresh

• After picking a key in the form, click Test connection: each endpoint is verified, and once all pass the platform's model list is fetched automatically into the form
• Platform card ⋯ menu:
  • API key connect / Sync models: verify auth → sync the model list
  • OAuth login (supported platforms): triggers the CLI login flow; models sync automatically on success
  • Settings (edit) and Delete

7.4 Command line

modelswap provider list              # list all providers (--json for scripts)
modelswap provider add               # add
modelswap provider delete <name>     # delete
modelswap provider auth              # auth status of all providers