Pre-release v1.10

Pre-Release v1.10

Setting this as pre-release because I don't have API keys for all the services needing more detailed testing.

Highlights:

• Dynamic model fetching for multiple providers
• OpenAI now uses the new Responses API
• Optional Base URL for OpenAI (proxies, regional endpoints)
• Updated Anthropic models and clearer provider naming
• Fully backward compatible with existing configurations

What’s new:

• OpenAI:
  • Models are now fetched dynamically from your endpoint.
  • Integration uses the OpenAI Responses API (/v1/responses) for improved
    compatibility.
  • New optional “Base URL” field:
    - Leave blank to use the official OpenAI endpoint.
    - Use a custom URL for proxies, regional endpoints, or compatible
    services.
  • “Custom…” option always available for manual model entry.
• Google Gemini:
  • Models are now fetched dynamically from Google.
  • “Custom…” option available if you need a specific model ID.
• Local OpenAI-Compatible Endpoints:
  • Previously labeled “OpenAI-Compatible” is now “Local OpenAI-Compatible” for
    clarity.
  • If your endpoint supports /v1/models, available models are fetched
    automatically.
  • Falls back to “Custom…” if dynamic listing is not supported.
• Anthropic (Claude):
  • Added support for latest models:
    - claude-opus-4-7
    - claude-sonnet-4-6
    - claude-haiku-4-5
• Provider and configuration improvements:
  • “local” provider refactored to “local_ollama” for clarity.
  • Improved configuration flows and validation.
  • Better error handling and logging (no user-facing breaking changes).
• Documentation:
  • Updated README with:
    - Dynamic model behavior
    - New Anthropic models
    - Base URL and Local OpenAI-Compatible details

Notes:

• Fully backward compatible:
  • Existing configurations continue to work without changes.
  • If a previously selected model is no longer available, choose a different one
    or use “Custom…” to specify it manually.