Feature Request: Integrate GPT-5.4 Codex support

[dmkononenko]

I would like to propose the integration of the GPT-5.4 Codex model into Kilo Code 5.x.x. As the ecosystem of LLMs continues to evolve, adding support for the latest iteration of Codex would significantly improve code generation accuracy, context handling, and overall developer productivity within the tool.

Proposed Changes:

• Add a new model entry for "GPT-5.4 Codex" in the model configuration settings.
• Update the API request structure to handle the specific requirements/parameters of the GPT-5.4 architecture.
• Ensure backward compatibility with existing model presets.

Why this is important:

• Performance: GPT-5.4 offers superior reasoning capabilities for complex refactoring and architecture tasks compared to current supported models.
• Reduced Latency: Better token efficiency could lead to faster completion speeds.
• Ecosystem parity: Staying up-to-date with the latest models is crucial for maintaining Kilo Code's competitive edge as a development assistant.

Additional Context:
I am currently using Kilo Code for development, and the ability to leverage the latest model versions would be a major game-changer.