docs: Document dynamic model fetching feature

[continue]

Summary

PR #12046 (merged 2026-04-08) added support for dynamically fetching provider models. This allows users to refresh model lists using their API key, with support for multiple providers including Ollama, OpenRouter, Anthropic, and Gemini.

What needs documentation

• Document the new "refresh models" button that appears after entering an API key
• Explain how dynamic model fetching works for different providers:
  • ollama: scrapes the library with icons
  • openrouter: hits public API on load
  • anthropic: calls /v1/models
  • gemini: calls v1beta/models
  • Others: use listModels() with optional API key/base
• Document that fetched models include contextLength, maxTokens, and supportsTools information
• Explain that capabilities are persisted to config.yaml when adding a model

New features to document

• Auto-loading for ollama and openrouter
• Refresh button for other providers after API key entry
• Model deduplication with hardcoded models (hardcoded take precedence)
• New model support: GPT-5.4 (pro, mini) and Gemma 4

Suggested locations to update

• docs/customize/model-roles/chat.mdx (already has a minor mention but needs expansion)
• Consider a new guide or section in model provider docs

Related PR

• feat: fetch provider models dynamically and fix isOpenSource pollution #12046

Generated with Continue

Co-Authored-By: Continue noreply@continue.dev
Co-authored-by: bekahhw rhawrot@gmail.com