Support for OpenAI's new GPT-4-1106-preview model, also known as GPT-4 Turbo with 128K context window.

[danielbowne]

Is this request related to a problem? Please describe.

I have been utilizing the 'Rubberduck: ChatGPT for Visual Studio Code' tool extensively in my work, leveraging its integration with OpenAI's GPT models for enhanced functionality. However, with the recent announcement of OpenAI's new GPT-4-1106-preview model, also known as GPT-4 Turbo, which offers an expanded 128K context window and lower operational costs. The absence of this model in 'Rubberduck: ChatGPT for Visual Studio Code' limits our ability to leverage the extended context for more complex queries and benefit from the cost efficiency that this new model promises.

Describe the solution you'd like

I propose that the 'Rubberduck: ChatGPT for Visual Studio Code' tool be updated to integrate the GPT-4-1106-preview model. This integration should include:

• Adaptation to the 128K context window: Ensuring that 'Rubberduck: ChatGPT for Visual Studio Code' can effectively utilize the extended context limit for more comprehensive and detailed interactions.
• Cost Optimization: Leveraging the lower cost of the GPT-4-1106-preview to make 'Rubberduck: ChatGPT for Visual Studio Code' more economical for users, particularly those who use the tool extensively.
• Maintaining Compatibility: Ensuring that the tool remains compatible with previous GPT models for users who prefer or require them.
• This update would significantly enhance 'Rubberduck: ChatGPT for Visual Studio Codes capabilities, allowing users to engage in more complex and extended dialogues with the AI, and reduce operational costs.

Additional context

As of now, there is no additional context or comparative tools to showcase. However, the integration of this model aligns with the forward-looking approach of keeping 'Rubberduck: ChatGPT for Visual Studio Code' at the forefront of AI-assisted tools. The extended context and cost-efficiency of GPT-4-1106-preview model are crucial for developers and professionals who rely on cutting-edge technology to streamline their workflows and optimize costs.

[restlessronin]

I notice that this support already seems to be in the modelfusion dependency, although the API has changed somewhat. This would be an important supporting feature for this PR.

@lgrammel, if the latest versions of modelfuision are stable, I can take a crack at updating rubberducky to use it.

[lgrammel]

@restlessronin yup, would be awesome to upgrade it. The changes to the modelfusion api should be minor. I also wanted to add ollama support at some point.

[lgrammel]

@restlessronin I've upgraded modelfusion to the latest version, so adding gpt-4-1106 support should be easy now