LLM EXPLORER 60,220 MODELS INDEXED

Qwen3.8 Flash Next NVFP4 FP8 PBWO DGX Spark by codegyver

By codegyver · 239 downloads

Qwen3.8 Flash Next NVFP4 FP8 PBWO DGX Spark is an open-source language model by codegyver. Features: 118.3b LLM, VRAM: 177.3GB, License: other, LLM Explorer Score: 0.32.

  8-bit Base model:inferact/qwen3.8-fl... Base model:quantized:inferact/...   Blackwell   Conversational   Dgx-spark   Endpoints compatible   Fp8   Fp8-pbwo   Image-text-to-text   Mixed-precision   Modelopt   Multimodal   Nvfp4   Qwen   Qwen3.8   Qwen3.8-flash-next   Qwen4 exp   Region:us   Safetensors   Sharded   Speculative-decoding   Tensorflow   Vllm

Qwen3.8 Flash Next NVFP4 FP8 PBWO DGX Spark Parameters and Internals

LLM NameQwen3.8 Flash Next NVFP4 FP8 PBWO DGX Spark
Repository πŸ€—https://huggingface.co/codegyver/Qwen3.8-Flash-Next-NVFP4-FP8-PBWO-DGX-Spark 
Base Model(s)  Inferact/Qwen3.8-Flash-Next-NVFP4   Inferact/Qwen3.8-Flash-Next-NVFP4
Model Size118.3b
Required VRAM177.3 GB
Updated2026-09-02
Maintainercodegyver
Model Typeqwen4_exp
Model Files  102.0 GB: 1-of-4   4.1 GB: 2-of-4   3.2 GB: 3-of-4   0.8 GB: 4-of-4   4.2 GB: 1-of-16   4.2 GB: 2-of-16   4.2 GB: 3-of-16   4.2 GB: 4-of-16   4.2 GB: 5-of-16   4.2 GB: 6-of-16   4.2 GB: 7-of-16   4.2 GB: 8-of-16   4.2 GB: 9-of-16   4.2 GB: 10-of-16   4.2 GB: 11-of-16   4.2 GB: 12-of-16   4.2 GB: 13-of-16   4.2 GB: 14-of-16   4.2 GB: 15-of-16   4.2 GB: 16-of-16   1.6 GB
Model ArchitectureQwen4ExpForConditionalGeneration
Licenseother
Model Max Length262144
Transformers Version5.8.0.dev0
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Errorsreplace

Best Alternatives to Qwen3.8 Flash Next NVFP4 FP8 PBWO DGX Spark

Best Alternatives
Context / RAM
Downloads
Likes
Qwen3.8 Flash Next NVFP40K / 180 GB35924
...lash Next Uncensored NVFP4 MTP0K / 1.4 GB3031
Note: green Score (e.g. "73.2") means that the model is better than codegyver/Qwen3.8-Flash-Next-NVFP4-FP8-PBWO-DGX-Spark.