Support for qwen3next architecture

[YAGP2024]

🥲 Failed to load the model

Failed to load model

error loading model: error loading model architecture: unknown model architecture: 'qwen3next'

lefromage/Qwen3-Next-80B-A3B-Instruct-GGUF

[aperepel]

FWIW, I've been using the qwen3-next-80b model for over a month in LM Studio.

I noticed, though, the arch is qwen3_next, not the qwen3next from your error message. This is probably a good starting point to dig in.

[YorkieDev]

@YAGP2024 will need to wait for support to be merged into llama.cpp main. The GGUF's you find for Qwen3 Next are made on a seperate fork and are not supported in LM Studio at this time. If you have a Mac, you can use the MLX versions of qwen3-next with the mlx runtimes. The llama.cpp runtimes do not have support for it yet.

[zeerd]

upstream is done, let's boom it !
ggml-org/llama.cpp#16095