[FR] Add support for multiple language models, including local models via Ollama, and GPT-4o-mini

[temperatio]

1–3 main use cases of the proposed feature:

As a user, I would like to choose between different AI language models, including self-hosted or locally run models like Ollama, to ensure flexibility and future-proof AI model integrations. This would allow AppFlowy-Cloud to support users who prefer running local models for privacy, or who want to switch between models from various providers.

What types of users can benefit from using your proposed feature:

• Users who prioritize data privacy and prefer to use local models.
• Developers and organizations looking for flexibility in AI model providers, especially in self-hosted environments.
• Users who want to continue using up-to-date models like GPT-4o-mini, as GPT-3.5-turbo will be deprecated next week, and there is no direct support for switching to an alternative model.

Additional context:

Given that GPT-3.5-turbo will be deprecated next week, the need for an easy transition to newer models such as GPT-4o-mini or alternatives from Anthropic or Ollama is critical. By providing users with a choice of models, AppFlowy-Cloud can ensure it remains adaptable and suitable for a variety of use cases, including self-hosted setups. Additionally, this feature could make it easier to switch between models as AI capabilities evolve or as certain models become deprecated or unavailable.