What are the hardware requirements for DeepSeek V4 Flash Nvfp4 DSpark?

DeepSeek V4 Flash Nvfp4 DSpark requires approximately 141.7 GB of VRAM and supports a context window of 1024K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed DeepSeek V4 Flash Nvfp4 DSpark and how large is it?

DeepSeek V4 Flash Nvfp4 DSpark is developed by nvidia, a model with 304.2b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate DeepSeek V4 Flash Nvfp4 DSpark?

DeepSeek V4 Flash Nvfp4 DSpark is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

DeepSeek V4 Flash Nvfp4 DSpark by nvidia — VRAM 141.7GB, 1024K context

Name: DeepSeek V4 Flash Nvfp4 DSpark
Author: nvidia

DeepSeek V4 Flash Nvfp4 DSpark is an open-source language model by nvidia. Features: 304.2b LLM, VRAM: 141.7GB, Context: 1024K, License: mit, LLM Explorer Score: 0.25.

Arxiv:2606.19348 8-bit Base model:deepseek-ai/deepsee... Base model:quantized:deepseek-... Conversational Deepseek v4 Deepseekv4 Dspark Fp8 Model optimizer Modelopt Nvfp4 Nvidia Quantized Region:us Safetensors Sharded Speculative-decoding Tensorflow

Model Card on HF 🤗: https://huggingface.co/nvidia/DeepSeek-V4-Flash-nvfp4-DSpark

DeepSeek V4 Flash Nvfp4 DSpark Parameters and Internals

LLM Name	DeepSeek V4 Flash Nvfp4 DSpark
Repository 🤗	https://huggingface.co/nvidia/DeepSeek-V4-Flash-nvfp4-DSpark
Base Model(s)	DeepSeek V4 Flash DeepSeek V4 Flash DSpark deepseek-ai/DeepSeek-V4-Flash deepseek-ai/DeepSeek-V4-Flash-DSpark
Model Size	304.2b
Required VRAM	141.7 GB
Updated	2026-08-15
Maintainer	nvidia
Model Type	deepseek_v4
Model Files	1.1 GB: 1-of-46 3.8 GB: 2-of-46 3.8 GB: 3-of-46 3.8 GB: 4-of-46 3.8 GB: 5-of-46 3.8 GB: 6-of-46 3.8 GB: 7-of-46 3.8 GB: 8-of-46 3.8 GB: 9-of-46 3.8 GB: 10-of-46 3.8 GB: 11-of-46 3.8 GB: 12-of-46 3.8 GB: 13-of-46 3.8 GB: 14-of-46 3.8 GB: 15-of-46 3.8 GB: 16-of-46 3.8 GB: 17-of-46 3.8 GB: 18-of-46 3.8 GB: 19-of-46 3.8 GB: 20-of-46 3.8 GB: 21-of-46 3.8 GB: 22-of-46 3.8 GB: 23-of-46 3.8 GB: 24-of-46 3.8 GB: 25-of-46 3.8 GB: 26-of-46 3.8 GB: 27-of-46 3.8 GB: 28-of-46 3.8 GB: 29-of-46 3.8 GB: 30-of-46 3.8 GB: 31-of-46 3.8 GB: 32-of-46 3.8 GB: 33-of-46 3.8 GB: 34-of-46 3.8 GB: 35-of-46 3.8 GB: 36-of-46 3.8 GB: 37-of-46 3.8 GB: 38-of-46
Model Architecture	DeepseekV4ForCausalLM
License	mit
Context Length	1048576
Model Max Length	1048576
Transformers Version	4.57.1
Vocabulary Size	129280
Torch Data Type	bfloat16

Rank the DeepSeek V4 Flash Nvfp4 DSpark capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to DeepSeek V4 Flash Nvfp4 DSpark

Best Alternatives	Context / RAM	Downloads	Likes
DeepSeek V4 Flash 0731	1024K / 152.3 GB	785771	2803
... V4 Flash 0731 Abliterated FP8	1024K / 145.1 GB	5149	35
DeepSeek V4 Flash 0731 NVFP4	1024K / 149.3 GB	7421	6
...Seek V4 Flash 0731 CRACK NVFP4	1024K / 141.5 GB	242	7
...4 Flash 0731 Abliterated NVFP4	1024K / 145.5 GB	952	11
DeepSeek V4 Flash 0731	1024K / 148.7 GB	89	25
DeepSeek V4 Flash 0731 NVFP4	1024K / 156.9 GB	75	2

Note: green Score (e.g. "73.2") means that the model is better than nvidia/DeepSeek-V4-Flash-nvfp4-DSpark.