MAI DS R1 by microsoft

 »  All LLMs  »  microsoft  »  MAI DS R1   URL Share it on

MAI DS R1 is an open-source language model by microsoft. Features: 671b LLM, VRAM: 171.7GB, Context: 160K, License: mit, LLM Explorer Score: 0.23.

Base model:deepseek-ai/deepsee... Base model:finetune:deepseek-a...   Conversational   Custom code   Deepseek v3   Endpoints compatible   Region:us   Safetensors   Sharded   Tensorflow

MAI DS R1 Benchmarks

nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

MAI DS R1 Parameters and Internals

LLM NameMAI DS R1
Repository 🤗https://huggingface.co/microsoft/MAI-DS-R1 
Base Model(s)  DeepSeek R1   deepseek-ai/DeepSeek-R1
Model Size671b
Required VRAM171.7 GB
Updated2026-03-27
Maintainermicrosoft
Model Typedeepseek_v3
Model Files  4.2 GB: 1-of-314   4.1 GB: 2-of-314   4.3 GB: 3-of-314   4.3 GB: 4-of-314   4.3 GB: 5-of-314   4.3 GB: 6-of-314   4.3 GB: 7-of-314   4.3 GB: 8-of-314   4.3 GB: 9-of-314   4.3 GB: 10-of-314   4.3 GB: 11-of-314   4.3 GB: 12-of-314   4.3 GB: 13-of-314   4.3 GB: 14-of-314   4.3 GB: 15-of-314   4.3 GB: 16-of-314   4.3 GB: 17-of-314   4.3 GB: 18-of-314   4.3 GB: 19-of-314   4.3 GB: 20-of-314   4.3 GB: 21-of-314   4.3 GB: 22-of-314   4.3 GB: 23-of-314   4.3 GB: 24-of-314   4.3 GB: 25-of-314   4.3 GB: 26-of-314   4.3 GB: 27-of-314   4.3 GB: 28-of-314   4.3 GB: 29-of-314   4.3 GB: 30-of-314   4.3 GB: 31-of-314   4.3 GB: 32-of-314   4.3 GB: 33-of-314   4.3 GB: 34-of-314   4.3 GB: 35-of-314   4.3 GB: 36-of-314   4.3 GB: 37-of-314   4.3 GB: 38-of-314   4.3 GB: 39-of-314   4.3 GB: 40-of-314
Model ArchitectureDeepseekV3ForCausalLM
Licensemit
Context Length163840
Model Max Length163840
Transformers Version4.46.3
Tokenizer ClassLlamaTokenizerFast
Beginning of Sentence Token<|begin▁of▁sentence|>
End of Sentence Token<|end▁of▁sentence|>
Vocabulary Size129280
Torch Data Typebfloat16

Quantized Models of the MAI DS R1

Model
Likes
Downloads
VRAM
MAI DS R1 GGUF613260 GB

Best Alternatives to MAI DS R1

Best Alternatives
Context / RAM
Downloads
Likes
Kimi K2 Thinking NVFP4256K / 210 GB482029
Kimi K2 Thinking BF16256K / 220 GB866
Kimi K2 Thinking256K / 383.2 GB304
DeepSeek V3.1160K / 180.4 GB150107819
DeepSeek V3.1 Terminus160K / 171.8 GB11934
DeepSeek V3.1 Base160K / 180.4 GB106441010
DeepSeek V3.1160K / 176.1 GB443
DeepSeek V3.0324160K / 376 GB358
DeepSeek R1160K / 358.8 GB7751
DeepSeek V3 Base160K / 171.8 GB23451684
Note: green Score (e.g. "73.2") means that the model is better than microsoft/MAI-DS-R1.

Rank the MAI DS R1 Capabilities

🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟

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  

What open-source LLMs or SLMs are you in search of? 52394 in total.

Our Social Media →  
Original data from HuggingFace, OpenCompass and various public git repos.
Check out Ag3ntum — our secure, self-hosted AI agent for server management.
Release v20260328a