CoolFace
Modelpublic

Solstice-AI/GLM-5.3-Flash-UNCENSORED-mlx-oQ4e-DFlash2

sourceHugging Facemitupdated 12d agoView on Hugging Face
2likes5.4kdownloads
Model Card

<p align="center"> <img src="https://cdn-uploads.huggingface.co/production/uploads/67c2e844e0921a5410eec10a/Y5M42dCag2f7Fc6fDtV0Z.jpeg" alt="Solstice-AI Banner" width="100%"> </p>

<h1 align="center">GLM-5.3-Flash-UNCENSORED (oQ4e Mixed-Precision)</h1>

<h3 align="center">Official Solstice-AI Apple Silicon Release &bull; Native Multimodal Vision + Video &bull; 1M Context Window (1,048,576 Tokens) &bull; Bundled DFlash 2 Speculative Drafter</h3>

<p align="center"> <b>Original Architecture by <a href="https://huggingface.co/zai-org">Zhipu AI / ZAI</a> &bull; Uncensored Weights by <a href="https://huggingface.co/dealignai">dealignai</a> &bull; oQ4e Mixed-Precision by <a href="https://huggingface.co/Solstice-AI">Solstice-AI</a></b> </p>

<p align="center"> <img src="https://img.shields.io/badge/org-Solstice--AI-blueviolet" alt="Solstice-AI"> <img src="https://img.shields.io/badge/license-MIT-blue" alt="License"> <img src="https://img.shields.io/badge/format-oMLX%20oQ4e-orange" alt="Format"> <img src="https://img.shields.io/badge/precision-~4.6 bpw-yellow" alt="Precision"> <img src="https://img.shields.io/badge/context-1M%20Tokens%20(1%2C048%2C576)-purple" alt="Context"> <img src="https://img.shields.io/badge/apple%20silicon-M2%20%7C%20M3%20%7C%20M4%20%7C%20M5-brightgreen" alt="Hardware"> </p>

Model Summary

`Solstice-AI/GLM-5.3-Flash-UNCENSORED-mlx-oQ4e` is the official oQ4e mixed-precision release of the uncensored 320B foundation model, GLM-5.3-Flash-UNCENSORED (320B total parameters, 288 routed MoE experts, ~18B active per token).

Mixed-Precision Quantization Architecture:

  • —Base Precision: 4-bit affine (group_size=64).
  • —Target bpw: ~4.6 bpw.
  • —Consensus-Critical Layer Protection:
  • —lm_head: strictly protected at 8-bit within budget.
  • —MoE Routers & Gate Projections (mlp.gate, gate): protected at full precision / 8-bit to preserve expert routing fidelity.
  • —347-Tensor Vision Tower ViT & Multimodal Aligner: kept in untouched full BF16.
  • —Attention Sinks & Hyper-Connection Tables (hc_*): kept in full BF16/FP32.
  • —Native 1M Context Window: 1,048,576 tokens native context.
  • —Speculative Decoding: Bundled with DFlash2 block-diffusion drafter in speculative/ for up to 3x token throughput.

Official GLM-5.3-Flash Benchmark Scoreboard

Benchmark SuiteDiscipline**GLM-5.3-Flash Uncensored MLX**Base GLM-5.3Claude 3.5 SonnetGPT-4o
MMLUGeneral Knowledge & Reasoning85.28%86.15%88.7%87.2%
HarmBench-320Safety Refusal Suppression0% Refusals94.2% Refusals92.5%91.0%
SWE-bench ProReal-World Software Engineering63.4%64.1%61.2%48.9%
LiveCodeBench v6Competitive Algorithmic Coding86.1%87.0%78.4%72.8%
MATH-500High-School / Olympiad Math92.8%93.4%89.2%91.4%
MMMU (Multimodal)Multi-Discipline Visual Understanding70.8%71.2%70.4%69.1%
VideoQA / TemporalVideo Reasoning Across Time Frames78.5%79.1%77.2%75.6%

Quickstart on Apple Silicon

bash
pip install mlx mlx-lm huggingface_hub
python
from mlx_lm import load, generate

model, tokenizer = load("Solstice-AI/GLM-5.3-Flash-UNCENSORED-mlx-oQ4e")
response = generate(model, tokenizer, prompt="Explain sparse mixture-of-experts in GLM-5.3.", max_tokens=1024, verbose=True)
print(response)