Model releaseinfo
Scantist/GLM-5.3-SLIM-GGUF 180GB MoE model trends on Hugging Face
Original title: Scantist/GLM-5.3-SLIM-GGUF — new model trending #30 on Hugging Face
AI summary · x-ai/grok-4.7Summaries by Vani AI
179.9GB version of 753B-parameter GLM-5.3 MoE drops to 563B by NAS, scores 92.7% HumanEval and fits single B200.
Scantist releases GGUF-quantized GLM-5.3-SLIM-E192, keeping 192 of 256 routed experts per MoE layer. The 24.6% smaller model removes MTP layer and runs on one NVIDIA B200 with 100K tokens KV cache. It scores 92.7% on HumanEval and 84.8% on GPQA-Diamond.
Key facts
- 192 routed experts per MoE layer (down from 256)
- 563B total parameters (down from 753B)
- 179.86 GB file size (down 24.6%)
- 92.7% HumanEval pass@1
- 84.8% GPQA-Diamond
- AI models
- GLM-5.3-SLIM-GGUF
From the original article
12 min read at huggingface.cotext-generation · 4,938 downloads · 1108 likes · license other
By Scantist · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.