Feature Request: GLM 4.5 MoE support

[jacekpoplawski]

Prerequisites

• I am running the latest code. Mention the version if possible as well.
• I carefully followed the README.md.
• I searched using keywords relevant to my issue to make sure that I am creating a new issue that is not already open (or closed).
• I reviewed the Discussions, and have a new and useful enhancement to share.

Feature Description

GLM 4.5 has been released
https://huggingface.co/collections/zai-org/glm-45-687c621d34bda8c9e4bf503b

it uses "Glm4MoeForCausalLM" architecture, llama.cpp currently supports only "Glm4ForCausalLM" and "Glm4vForConditionalGeneration"

Motivation

it's a SOTA MoE model

Possible Implementation

vllm-project/vllm#20736

[olumolu]

Please provide support

[Fristender]

Also MTP support would significantly increase token generation speed. See #13236.

[Kreijstal]

finally claude sonnet @ home

[nimishchaudhari]

Does someone have a working chat template for compatibility with some mainstream app like cline? I am having troubles with unsloth quants I downloaded. Waiting to unleash Claude at home for vibe coding xd

[ddh0]

@nimishchaudhari Please see #15904