LLM EXPLORER 60,645 MODELS INDEXED

Qwen 3.8 Next 40B Exp MoE Pruned Bf16 by osk-arr00

By osk-arr00 · 612 downloads
HF Repo ↗ Copy URL

Qwen 3.8 Next 40B Exp MoE Pruned Bf16 is an open-source language model by osk-arr00. Features: 43b LLM, VRAM: 32.9GB, License: apache-2.0, MoE, LLM Explorer Score: 0.32.

Base model:davd-b01/qwen-3.8-n... Base model:finetune:davd-b01/q...   Conversational   Endpoints compatible   Experimental   Image-text-to-text   Moe   Not-for-production   Pruning   Qwen4 exp   Region:us   Research   Safetensors   Sharded   Tensorflow   Testbed
Model Card on HF 🤗: https://huggingface.co/Davd-b01/qwen-3.8-next-40b-exp-moe-pruned-bf16 

Qwen 3.8 Next 40B Exp MoE Pruned Bf16 Parameters and Internals

LLM NameQwen 3.8 Next 40B Exp MoE Pruned Bf16
Repository 🤗https://huggingface.co/Davd-b01/qwen-3.8-next-40b-exp-moe-pruned-bf16 
Base Model(s)  osk-arr00/qwen-3.8-next-40b-exp-clean-bf16   osk-arr00/qwen-3.8-next-40b-exp-clean-bf16
Model Size43b
Required VRAM32.9 GB
Updated2026-09-07
Maintainerosk-arr00
Model Typeqwen4_exp
Model Files  1.0 GB: 1-of-131   0.8 GB: 2-of-131   0.6 GB: 3-of-131   0.8 GB: 4-of-131   2.1 GB: 5-of-131   1.7 GB: 37-of-131   0.8 GB: 38-of-131   0.5 GB: 39-of-131   0.8 GB: 40-of-131   0.7 GB: 41-of-131   0.8 GB: 42-of-131   0.6 GB: 43-of-131   0.8 GB: 44-of-131   0.6 GB: 45-of-131   0.8 GB: 46-of-131   0.5 GB: 47-of-131   0.8 GB: 48-of-131   0.7 GB: 49-of-131   0.8 GB: 50-of-131   0.6 GB: 51-of-131   0.8 GB: 52-of-131   0.6 GB: 53-of-131   0.8 GB: 54-of-131   0.5 GB: 55-of-131   0.8 GB: 56-of-131   0.7 GB: 57-of-131   0.8 GB: 58-of-131   0.5 GB: 59-of-131   1.0 GB: 60-of-131   0.8 GB: 61-of-131   1.0 GB: 62-of-131   0.8 GB: 63-of-131   0.6 GB: 64-of-131   0.8 GB: 65-of-131   0.5 GB: 66-of-131   0.8 GB: 67-of-131   0.7 GB: 68-of-131   0.8 GB: 69-of-131   0.6 GB: 70-of-131   0.8 GB: 71-of-131   0.6 GB: 72-of-131   0.8 GB: 73-of-131
Model ArchitectureQwen4ExpForConditionalGeneration
Licenseapache-2.0
Transformers Version5.8.0.dev0