A quantized version of the GLM-5.3 model, named GLM-5.3-UNCENSORED EXL3 3.0bpw, has been released on Hugging Face. This model is based on the dealignai/GLM-5.3-UNCENSORED-FP8, which itself is a weight-edited variant of zai-org/GLM-5.3. The new release utilizes ExLlamaV3 quantization and features a 753B parameter architecture with 256 routed experts. Initial evaluations suggest minimal degradation in performance compared to the original FP8 source, though a caveat regarding tool calling argument parsing has been noted. AI
IMPACT Provides a more accessible, quantized version of a large model for researchers and developers.
RANK_REASON Release of a quantized version of an existing model, not a novel frontier model release.
Read on Hugging Face Trending Models →
- dealignai/GLM-5.3-UNCENSORED-FP8
- ExLlamaV3
- GLM-5.3-UNCENSORED EXL3 3.0bpw
- GPT-4.1
- Infatoshi/GLM-5.3-UNCENSORED-EXL3-3.0bpw
- OpenRouter
- TabbyAPI
- zai-org/GLM-5.3
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →