LLM EXPLORER 61,755 MODELS INDEXED

Qwen3.8 Flash Next EXL3 2.50bpw by r0b0tlab

By r0b0tlab · 97 downloads
HF Repo ↗ Copy URL

Qwen3.8 Flash Next EXL3 2.50bpw is an open-source language model by r0b0tlab. Features: 22.3b LLM, VRAM: 44.5GB, License: apache-2.0, LLM Explorer Score: 0.32.

  2-bit Base model:quantized:qwen/qwen... Base model:qwen/qwen3.8-flash-...   Conversational   Exl3   Exllamav3   Image-text-to-text   Moe   Mtp   Quantization   Qwen4 exp   Region:us   Rtx3090   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/r0b0tlab/Qwen3.8-Flash-Next-EXL3-2.50bpw 

Qwen3.8 Flash Next EXL3 2.50bpw Parameters and Internals

LLM NameQwen3.8 Flash Next EXL3 2.50bpw
Repository 🤗https://huggingface.co/r0b0tlab/Qwen3.8-Flash-Next-EXL3-2.50bpw 
Base Model(s)  Qwen/Qwen3.8-Flash-Next   Qwen/Qwen3.8-Flash-Next
Model Size22.3b
Required VRAM44.5 GB
Updated2026-09-20
Maintainerr0b0tlab
Model Typeqwen4_exp
Model Files  8.4 GB: 1-of-6   8.6 GB: 2-of-6   8.4 GB: 3-of-6   7.6 GB: 4-of-6   8.1 GB: 5-of-6   3.4 GB: 6-of-6   19.8 GB
Model ArchitectureQwen4ExpForConditionalGeneration
Licenseapache-2.0
Model Max Length262144
Transformers Version5.8.0.dev0
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Errorsreplace

Best Alternatives to Qwen3.8 Flash Next EXL3 2.50bpw

Best Alternatives
Context / RAM
Downloads
Likes
... Next Abliterated EXL3 2.50bpw0K / 44.5 GB5602
Note: green Score (e.g. "73.2") means that the model is better than r0b0tlab/Qwen3.8-Flash-Next-EXL3-2.50bpw.