[Bug] MLX version of Gemma 4 26B fails to load: "Model type gemma4 not supported" (macOS)

**Which version of LM Studio?** LM Studio 0.4.8+1 (Note: Using llama.cpp core 2.10.1) **Which operating system?** macOS (26.4) - Apple Silicon M1 Max (64GB RAM) **What is the bug?** When attempting to load an MLX-formatted Gemma 4 model (specifically gemma-4-26B-A4B from the mlx-community), LM Studio throws a ValueError and aborts. The error indicates that the underlying MLX backend (mlx_vlm) bundled with LM Studio does not yet have the definitions for the gemma4 architecture. Just like (#1728) the recent llama.cpp backend bump required to fix the GGUF loading for Gemma 4, it appears the MLX engine dependency also needs an upstream bump to support this new release. **Screenshots** <img width="443" height="170" alt="Image" src="https://github.com/user-attachments/assets/b5fc249f-d684-4e35-b14b-68363a580ec7" /> **Logs** ``` Error when loading model: ValueError: Model type gemma4 not supported. Error: No module named 'mlx_vlm.models.gemma4' ``` **To Reproduce** Steps to reproduce the behavior: 1. Download the MLX version of Gemma 4 (e.g., gemma-4-26B-A4B from mlx-community). 2. Navigate to the Chat interface. 3. Select the MLX model from the top drop-down menu to load it. 4. The loading process instantly fails and the UI throws the ValueError regarding the unsupported gemma4 module.

Posted by ranajahanzaib on Apr 3, 2026, 10:26 AM

Tags

• reliability (Software)

Related issues

• [BUG] Cowork (Windows): all Plan9 shares fail with "Plan9 mount failed: invalid argument" after Windows update KB5124008 (26200.9445) — uninstalling the KB fixe
• Feature Request: Auto-resume CLI session when usage limit resets
• Please deobfuscate the Python
• # 233 Developers Discussed Abandoning the Office Add-in Platform — An Open Letter on the Office.js Stability, Security & Trust Crisis
• [FEATURE] Allow users to configure which shell the Bash tool uses or inherit initial shell
• Admin Materials and Surfaces