LM Studio + diffusion_gemma

[salamanders]

During model load of https://huggingface.co/mlx-community/diffusiongemma-26B-A4B-it-6bit

🥲 Failed to load the model

Failed to load model.

Error when loading model: ValueError: Model type diffusion_gemma not supported. Error: No module named 'mlx_vlm.speculative.drafters.diffusion_gemma'

LM Studio MLX
v1.8.5
Apple MLX engine, based on the MLX Python implementation
Latest version

Metal llama.cpp
v2.21.0
Apple Metal accelerated llama.cpp engine
Latest version

Harmony (Mac)
v0.3.5
Chat history renderer and parser from OpenAI
Latest version

[taills]

https://huggingface.co/unsloth/diffusiongemma-26B-A4B-it-GGUF

2026-06-12 02:13:41 [DEBUG]
 LlamaV4::load called with model path: /Users/admin/.lmstudio/models/unsloth/diffusiongemma-26B-A4B-it-GGUF/diffusiongemma-26B-A4B-it-Q4_K_M.gguf
LlamaV4::load config: n_parallel=4 n_ctx=8192 kv_unified=true
2026-06-12 02:13:41 [DEBUG]
 0.00.106.991 I srv    load_model: loading model '/Users/admin/.lmstudio/models/unsloth/diffusiongemma-26B-A4B-it-GGUF/diffusiongemma-26B-A4B-it-Q4_K_M.gguf'
2026-06-12 02:13:41 [DEBUG]
 0.00.272.377 E llama_model_load: error loading model: unknown model architecture: 'diffusion-gemma'
2026-06-12 02:13:41 [DEBUG]
 0.00.272.510 E llama_model_load_from_file_impl: failed to load model
0.00.272.512 E common_init_from_params: failed to load model '/Users/admin/.lmstudio/models/unsloth/diffusiongemma-26B-A4B-it-GGUF/diffusiongemma-26B-A4B-it-Q4_K_M.gguf'
0.00.272.513 E srv    load_model: failed to load model, '/Users/admin/.lmstudio/models/unsloth/diffusiongemma-26B-A4B-it-GGUF/diffusiongemma-26B-A4B-it-Q4_K_M.gguf': error loading model: unknown model architecture: 'diffusion-gemma'
2026-06-12 02:13:41 [DEBUG]
 [LLMProcess] Failed to load model _0x105a04 [Error]: Failed to load model.
    at _0x1dfaf9.loadModel (/Applications/LM Studio.app/Contents/Resources/app/.webpack/lib/llmworker.js:1:613793)
    at process.processTicksAndRejections (node:internal/process/task_queues:104:5)
    at async _0x1dfaf9.handleMessage (/Applications/LM Studio.app/Contents/Resources/app/.webpack/lib/llmworker.js:1:605893) {
  cause: 'Failed to load model',
  suggestion: undefined,
  errorData: undefined,
  data: undefined,
  displayData: undefined,
  title: 'Failed to load model.'
}

[michal-kindl]

I have a similar issue, apparently diffusiongemma not yet supported, snippet from logs:

0.00.898.136 E llama_model_load: error loading model: unknown model architecture: 'diffusion-gemma'
0.00.898.145 E llama_model_load_from_file_impl: failed to load model
0.00.898.147 E common_init_from_params: failed to load model 'D:\Michal\_LM_Studio_Data\unsloth\diffusiongemma-26B-A4B-it-GGUF\diffusiongemma-26B-A4B-it-Q4_K_M.gguf'
0.00.898.155 E srv    load_model: failed to load model, 'D:\Michal\_LM_Studio_Data\unsloth\diffusiongemma-26B-A4B-it-GGUF\diffusiongemma-26B-A4B-it-Q4_K_M.gguf': error loading model: unknown model architecture: 'diffusion-gemma'

HW & environment:
OS: Windows 11 Home, x64, version 25H2, build 26200.8655
LM Studio: version 0.4.17 (Build 2),
Engine: CUDA llama.cpp (Windows) v2.22.0
Processor: Intel(R) Core(TM) i7-6700K CPU @ 4.00GHz (4.00 GHz)
RAM: 64,0 GB
GPUs:
NVIDIA GeForce RTX 4060 Ti (16 GB)
NVIDIA GeForce GTX 1060 6GB (6 GB)

[Jakeshadow]

LM Studio's MLX engine doesn't support diffusion_gemma yet
unfortunately. I hit the same wall.

Two options that work right now while waiting for native support:

vLLM (fastest, but needs NVIDIA GPU):
pip install vllm --pre
python -c "from vllm import LLM; llm = LLM('google/diffusiongemma-26B-A4B-it'); print(llm.generate('Hello'))"

llama.cpp PR branch (works on anything, including Mac):
git clone -b pr/24427 https://github.com/ggerganov/llama.cpp.git
cd llama.cpp && make -j
./llama-cli -m diffusiongemma-Q4_K_M.gguf -p "Hello"

Both give 200-1000+ tok/s depending on GPU. I've got a full
comparison of all the methods that actually work here:
https://diffusiongemma.dev/

Fair warning: Apple Silicon doesn't get the speedup NVIDIA does
with this model — it really wants CUDA tensor cores.