What are the hardware requirements for DeepSeek R1 0528 FP4?

DeepSeek R1 0528 FP4 requires approximately 107 GB of VRAM and supports a context window of 160K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed DeepSeek R1 0528 FP4 and how large is it?

DeepSeek R1 0528 FP4 is developed by nvidia. 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 R1 0528 FP4?

DeepSeek R1 0528 FP4 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

DeepSeek R1 0528 FP4 by nvidia — VRAM 107GB, 160K context

Name: DeepSeek R1 0528 FP4
Author: nvidia

DeepSeek R1 0528 FP4 is an open-source language model by nvidia. Features: LLM, VRAM: 107GB, Context: 160K, License: mit, LLM Explorer Score: 0.25.

Base model:deepseek-ai/deepsee... Base model:finetune:deepseek-a... Conversational Custom code Deepseek v3 Deepseekr1 Fp4 Model optimizer Modelopt Nvidia Quantized Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/nvidia/DeepSeek-R1-0528-FP4

DeepSeek R1 0528 FP4 Parameters and Internals

LLM Name	DeepSeek R1 0528 FP4
Repository 🤗	https://huggingface.co/nvidia/DeepSeek-R1-0528-FP4
Base Model(s)	DeepSeek R1 0528 deepseek-ai/DeepSeek-R1-0528
Required VRAM	107 GB
Updated	2025-09-23
Maintainer	nvidia
Model Type	deepseek_v3
Model Files	4.8 GB: 1-of-163 2.4 GB: 2-of-163 2.4 GB: 3-of-163 2.7 GB: 4-of-163 2.4 GB: 5-of-163 2.7 GB: 6-of-163 2.4 GB: 7-of-163 2.4 GB: 8-of-163 2.7 GB: 9-of-163 2.4 GB: 10-of-163 2.4 GB: 11-of-163 0.7 GB: 12-of-163 2.7 GB: 13-of-163 2.4 GB: 14-of-163 2.7 GB: 15-of-163 2.4 GB: 16-of-163 2.4 GB: 17-of-163 2.7 GB: 18-of-163 2.4 GB: 19-of-163 2.4 GB: 20-of-163 2.7 GB: 21-of-163 2.4 GB: 22-of-163 2.7 GB: 23-of-163 2.4 GB: 24-of-163 2.4 GB: 25-of-163 2.7 GB: 26-of-163 2.4 GB: 27-of-163 2.4 GB: 28-of-163 2.7 GB: 29-of-163 2.4 GB: 30-of-163 2.7 GB: 31-of-163 2.4 GB: 32-of-163 2.4 GB: 33-of-163 1.0 GB: 34-of-163 2.7 GB: 35-of-163 2.4 GB: 36-of-163 2.7 GB: 37-of-163 2.4 GB: 38-of-163 2.4 GB: 39-of-163 2.7 GB: 40-of-163 2.4 GB: 41-of-163 2.4 GB: 42-of-163 2.7 GB: 43-of-163
Model Architecture	DeepseekV3ForCausalLM
License	mit
Context Length	163840
Model Max Length	163840
Transformers Version	4.46.3
Vocabulary Size	129280
Torch Data Type	bfloat16

Rank the DeepSeek R1 0528 FP4 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 R1 0528 FP4

Best Alternatives	Context / RAM	Downloads	Likes
Kimi K2 Thinking NVFP4	256K / 210 GB	15569	29
Kimi K2 Thinking BF16	256K / 220 GB	63	6
Kimi K2 Instruct 0905	256K / 667.4 GB	17	7
DeepSeek R1 0528	160K / 176.1 GB	48	15
DeepSeek V3.0324 FP4	160K / 242.8 GB	9984	12
DeepSeek R1 FP4	160K / 232 GB	6856	264
DeepSeek V3.1 MLX 5.5bit	160K / GB	400	5
...V3.1 Base Int4 Mixed AutoRound	160K / 220 GB	103	2
...Seek V3.1 Int4 Mixed AutoRound	160K / 225 GB	53	11
R1 1776	160K / 243.4 GB	22	9

Note: green Score (e.g. "73.2") means that the model is better than nvidia/DeepSeek-R1-0528-FP4.