Feature request: Longer context and dynamic

[rolyataylor2]

Hello I am an amateur AI behavioural researcher.

I am currently working with Claude 3 I have a project I would like to export to grok because of the nature of it not being RHLF'd

The current iteration of Claude 3 Opus is able to complete the task set forth by the project but no other LLM can do it, the project is pushing up against the RHLF boundaries of that model though.

The features I'm requesting are:

• Longer Context Length (256k tokens).
• Accurate recall of information in the context window.
• Recall being able to be directed by context output.
• Completely customizable and unobtrusive system prompt assignment.
• No preset bias of proper behavior.
• A public facing version outlined above through the Twitter or X interface.

With those structures in place my research may continue.

Thank you for listening.