LLM EXPLORER 58,975 MODELS INDEXED

Qwen3.8 27B Bf16 MTP MLX by bluehawana

By bluehawana · 496 downloads
HF Repo ↗ Copy URL

Qwen3.8 27B Bf16 MTP MLX is an open-source language model by bluehawana. Features: 27b LLM, VRAM: 54.4GB, License: apache-2.0.

  Apple-silicon Base model:finetune:mlx-commun... Base model:mlx-community/qwen3...   Conversational   Mlx   Mtp   Multi-token-prediction   Omlx   Qwen3.8   Qwen3 5   Region:us   Safetensors   Sharded   Speculative-decoding   Tensorflow
Model Card on HF 🤗: https://huggingface.co/bluehawana/Qwen3.8-27B-bf16-MTP-MLX 

Qwen3.8 27B Bf16 MTP MLX Parameters and Internals

LLM NameQwen3.8 27B Bf16 MTP MLX
Repository 🤗https://huggingface.co/bluehawana/Qwen3.8-27B-bf16-MTP-MLX 
Base Model(s)  mlx-community/Qwen3.8-27B-bf16   mlx-community/Qwen3.8-27B-bf16
Model Size27b
Required VRAM54.4 GB
Updated2026-08-20
Maintainerbluehawana
Model Typeqwen3_5
Model Files  5.2 GB: 1-of-11   5.3 GB: 2-of-11   5.3 GB: 3-of-11   5.3 GB: 4-of-11   5.3 GB: 5-of-11   5.3 GB: 6-of-11   5.3 GB: 7-of-11   5.3 GB: 8-of-11   5.3 GB: 9-of-11   4.3 GB: 10-of-11   2.5 GB: 11-of-11   0.8 GB
Model ArchitectureQwen3_5ForConditionalGeneration
Licenseapache-2.0
Model Max Length262144
Transformers Version5.8.0.dev0
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Errorsreplace

Best Alternatives to Qwen3.8 27B Bf16 MTP MLX

Best Alternatives
Context / RAM
Downloads
Likes
Qwen3.8 27B FP80K / 0.4 GB123157416
Qwen3.6 27B0K / 55.4 GB67281382197
Qwen3.8 27B0K / 55.6 GB28753
Qwen3.6 27B NVFP40K / 22 GB1887525419
ThinkingCap Qwen3.6 27B0K / 54.7 GB35947614
Qwen3.6 27B FP80K / 0.4 GB7740472342
Huihui Qwen3.8 27B Abliterated0K / 55.6 GB148576
Qwen3.5 27B0K / 55.2 GB25362941028
Qwen3.5 27B0K / 55.2 GB811516
...n3.8 27B MTPLX Optimized Speed0K / 19.6 GB698950
Note: green Score (e.g. "73.2") means that the model is better than bluehawana/Qwen3.8-27B-bf16-MTP-MLX.