LLM EXPLORER 58,133 MODELS INDEXED

DeepSeek V4 Flash 0731 OptiQ 2bit by mlx-community

By mlx-community · 983 downloads
HF Repo ↗ Copy URL

DeepSeek V4 Flash 0731 OptiQ 2bit is an open-source language model by mlx-community. Features: 25.5b LLM, VRAM: 92.6GB, Context: 1024K, License: mit, Quantized, LLM Explorer Score: 0.33.

  2-bit   2bit   Apple-silicon Base model:deepseek-ai/deepsee... Base model:quantized:deepseek-...   Conversational   Deepseek   Deepseek-v4   Deepseek v4   Mixed-precision   Mlx   Moe   Optiq   Quantized   Region:us   Safetensors   Sharded   Ssd-streaming   Tensorflow
Model Card on HF 🤗: https://huggingface.co/mlx-community/DeepSeek-V4-Flash-0731-OptiQ-2bit 

DeepSeek V4 Flash 0731 OptiQ 2bit Parameters and Internals

LLM NameDeepSeek V4 Flash 0731 OptiQ 2bit
Repository 🤗https://huggingface.co/mlx-community/DeepSeek-V4-Flash-0731-OptiQ-2bit 
Base Model(s)  DeepSeek V4 Flash 0731   deepseek-ai/DeepSeek-V4-Flash-0731
Model Size25.5b
Required VRAM92.6 GB
Updated2026-08-11
Maintainermlx-community
Model Typedeepseek_v4
Model Files  2.4 GB: 1-of-42   2.5 GB: 2-of-42   2.0 GB: 3-of-42   2.0 GB: 4-of-42   2.5 GB: 5-of-42   2.0 GB: 6-of-42   2.5 GB: 7-of-42   2.0 GB: 8-of-42   2.5 GB: 9-of-42   2.0 GB: 10-of-42   2.0 GB: 11-of-42   2.0 GB: 12-of-42   2.5 GB: 13-of-42   2.0 GB: 14-of-42   2.5 GB: 15-of-42   2.0 GB: 16-of-42   2.5 GB: 17-of-42   2.0 GB: 18-of-42   2.0 GB: 19-of-42   2.5 GB: 20-of-42   2.5 GB: 21-of-42   2.0 GB: 22-of-42   2.5 GB: 23-of-42   2.0 GB: 24-of-42   2.0 GB: 25-of-42   2.5 GB: 26-of-42   2.0 GB: 27-of-42   2.5 GB: 28-of-42   2.0 GB: 29-of-42   2.5 GB: 30-of-42   2.0 GB: 31-of-42   2.0 GB: 32-of-42   2.0 GB: 33-of-42   2.5 GB: 34-of-42   2.0 GB: 35-of-42   2.5 GB: 36-of-42   2.0 GB: 37-of-42   2.5 GB: 38-of-42   2.7 GB: 39-of-42   2.7 GB: 40-of-42   2.7 GB: 41-of-42   0.1 GB: 42-of-42
Quantization Type2bit
Model ArchitectureDeepseekV4ForCausalLM
Licensemit
Context Length1048576
Model Max Length1048576
Transformers Version4.57.1
Vocabulary Size129280
Torch Data Typebfloat16