LM Studio hack unlocks Gemma 4 MLX

The fact that Gemma 4 support is already bundled but blocked suggests LM Studio is polishing the integration before a full release, or awaiting final stability patches for its MLX engine implementation.

• –Modifying `generate.py` bypasses a `ValueError` that prevents Gemma 4 from loading despite the underlying engine support being present.
• –Successful community tests on M4 Pro hardware confirm that the bundled `mlx-vlm` and `mlx-lm` libraries are ready for the new model architecture.
• –Gemma 4 represents a major multimodal release for Google, featuring a 256K context window and native support for audio, image, and video inputs.
• –While GGUF versions are already functional, this hack provides a path for users preferring MLX's specific performance profile on Mac.
• –Caution is advised as direct edits to vendor extensions can be overwritten by future app updates or cause unexpected stability issues.