LLM EXPLORER 60,277 MODELS INDEXED

DeepSeek R1 2bit by mlx-community

By mlx-community · 167 downloads
HF Repo ↗ Copy URL

DeepSeek R1 2bit is an open-source language model by mlx-community. Features: 84b LLM, VRAM: 238.3GB, Context: 160K, License: mit, Quantized, LLM Explorer Score: 0.17.

  2-bit   2bit Base model:deepseek-ai/deepsee... Base model:quantized:deepseek-...   Conversational   Custom code   Deepseek v3   Endpoints compatible   Mlx   Quantized   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/mlx-community/DeepSeek-R1-2bit 

DeepSeek R1 2bit Parameters and Internals

LLM NameDeepSeek R1 2bit
Repository 🤗https://huggingface.co/mlx-community/DeepSeek-R1-2bit 
Base Model(s)  DeepSeek R1   deepseek-ai/DeepSeek-R1
Model Size84b
Required VRAM238.3 GB
Updated2026-08-03
Maintainermlx-community
Model Typedeepseek_v3
Model Files  5.3 GB: 1-of-50   5.3 GB: 2-of-50   4.8 GB: 3-of-50   5.3 GB: 4-of-50   4.8 GB: 5-of-50   5.3 GB: 6-of-50   4.9 GB: 7-of-50   5.3 GB: 8-of-50   4.8 GB: 9-of-50   5.3 GB: 10-of-50   4.8 GB: 11-of-50   5.3 GB: 12-of-50   4.9 GB: 13-of-50   5.3 GB: 14-of-50   4.8 GB: 15-of-50   5.3 GB: 16-of-50   4.8 GB: 17-of-50   5.3 GB: 18-of-50   4.9 GB: 19-of-50   5.3 GB: 20-of-50   4.8 GB: 21-of-50   5.3 GB: 22-of-50   4.8 GB: 23-of-50   5.3 GB: 24-of-50   4.9 GB: 25-of-50   5.3 GB: 26-of-50   4.8 GB: 27-of-50   5.3 GB: 28-of-50   4.8 GB: 29-of-50   5.3 GB: 30-of-50   4.9 GB: 31-of-50   5.3 GB: 32-of-50   4.8 GB: 33-of-50   5.3 GB: 34-of-50   4.8 GB: 35-of-50   5.3 GB: 36-of-50   4.9 GB: 37-of-50   5.3 GB: 38-of-50   4.8 GB: 39-of-50   5.3 GB: 40-of-50   4.8 GB: 41-of-50   5.3 GB: 42-of-50   4.9 GB: 43-of-50   5.3 GB: 44-of-50   4.8 GB: 45-of-50   5.3 GB: 46-of-50   4.8 GB: 47-of-50
Quantization Type2bit
Model ArchitectureDeepseekV3ForCausalLM
Licensemit
Context Length163840
Model Max Length163840
Transformers Version4.46.3
Vocabulary Size129280
Torch Data Typebfloat16