I created a 140 GB IQ2_XXS REAP quant of GLM 5.2 for coding. Looking for testers.

GLM-5.2-504B-Code-GGUF is an imatrix calibrated quant of the most liked REAP, which is 0xSero/GLM-5.2-REAP-504B-GGUF .

I also uploaded the imatrix file and the modified llama-quant.cpp so you can compile llama.cpp and create your own quant.

I wonder how it compares to smaller LLMs like Qwen 3.6 27b and DeepSeek Flash v4.

https://huggingface.co/icecubetr/GLM-5.2-504B-Code-GGUF