LLM EXPLORER 59,516 MODELS INDEXED

Qwen3.8 2B Distill Mlx 5Bit by keXjos

By keXjos · 387 downloads
HF Repo ↗ Copy URL

Qwen3.8 2B Distill Mlx 5Bit is an open-source language model by keXjos. Features: 2b LLM, VRAM: 1.3GB, License: apache-2.0, Quantized, LLM Explorer Score: 0.33.

  5-bit   5bit Base model:empero-ai/qwen3.8-2... Base model:quantized:empero-ai...   Conversational   Distillation   Edge   Empero-ai   En   Endpoints compatible   Function-calling   Image-text-to-text   Mlx   Mlx-my-repo   Quantized   Qwen3.5   Qwen3.8   Qwen3 5   Reasoning   Region:us   Safetensors   Sft
Model Card on HF 🤗: https://huggingface.co/keXjos/Qwen3.8-2B-Distill-mlx-5Bit 

Qwen3.8 2B Distill Mlx 5Bit Parameters and Internals

LLM NameQwen3.8 2B Distill Mlx 5Bit
Repository 🤗https://huggingface.co/keXjos/Qwen3.8-2B-Distill-mlx-5Bit 
Base Model(s)  empero-ai/Qwen3.8-2B-Distill   empero-ai/Qwen3.8-2B-Distill
Model Size2b
Required VRAM1.3 GB
Updated2026-08-25
MaintainerkeXjos
Model Typeqwen3_5
Model Files  1.3 GB
Supported Languagesen
Quantization Type5bit
Model ArchitectureQwen3_5ForConditionalGeneration
Licenseapache-2.0
Model Max Length262144
Transformers Version4.57.0.dev0
Tokenizer ClassTokenizersBackend
Padding Token<|endoftext|>
Errorsreplace

Best Alternatives to Qwen3.8 2B Distill Mlx 5Bit

Best Alternatives
Context / RAM
Downloads
Likes
Qwen3.5 2B MLX 4bit0K / 1.7 GB123102
Qwen3.5 2B 6bit0K / 2.2 GB86807
Qwen3.5 2B 4bit0K / 1.7 GB53745
Spartan 1.2 2B MLX 8bit0K / 1.1 GB270
Qwen3.5 2B OptiQ 4bit0K / 0.7 GB129612
Qwen3.5 2B 8bit0K / 2.7 GB2021
Qwen3.5 2B0K / 4.5 GB2633311352
Qwen3.8 2B Distill0K / 4.5 GB205624
Qwen3.5 2B Base0K / 4.5 GB80959182
Qwen3.8 2B0K / 4.5 GB31913
Note: green Score (e.g. "73.2") means that the model is better than keXjos/Qwen3.8-2B-Distill-mlx-5Bit.