LLM EXPLORER 58,608 MODELS INDEXED

Kimi K2.6 Mlx DQ3 K M Q8 by mlx-community

By mlx-community · 6157 downloads
HF Repo ↗ Copy URL

Kimi K2.6 Mlx DQ3 K M Q8 is an open-source language model by mlx-community. Features: 1026.4b LLM, VRAM: 186GB, License: other, Quantized, LLM Explorer Score: 0.29.

  Arxiv:2505.02390   4-bit Base model:moonshotai/kimi-k2.... Base model:quantized:moonshota...   Conversational   Custom code   Feature-extraction   Kimi k25   Mlx   Q8   Quantized   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/mlx-community/Kimi-K2.6-mlx-DQ3_K_M-q8 

Kimi K2.6 Mlx DQ3 K M Q8 Parameters and Internals

LLM NameKimi K2.6 Mlx DQ3 K M Q8
Repository 🤗https://huggingface.co/mlx-community/Kimi-K2.6-mlx-DQ3_K_M-q8 
Base Model(s)  moonshotai/Kimi-K2.6   moonshotai/Kimi-K2.6
Model Size1026.4b
Required VRAM186 GB
Updated2026-08-04
Maintainermlx-community
Model Typekimi_k25
Model Files  4.3 GB: 1-of-99   2.5 GB: 2-of-99   4.0 GB: 3-of-99   4.9 GB: 4-of-99   4.0 GB: 5-of-99   4.9 GB: 6-of-99   4.0 GB: 7-of-99   4.9 GB: 8-of-99   4.0 GB: 9-of-99   4.9 GB: 10-of-99   3.3 GB: 11-of-99   4.9 GB: 12-of-99   5.1 GB: 13-of-99   5.1 GB: 14-of-99   4.9 GB: 15-of-99   5.1 GB: 16-of-99   5.1 GB: 17-of-99   4.9 GB: 18-of-99   3.3 GB: 19-of-99   4.9 GB: 20-of-99   5.1 GB: 21-of-99   5.1 GB: 22-of-99   4.9 GB: 23-of-99   5.1 GB: 24-of-99   5.1 GB: 25-of-99   4.9 GB: 26-of-99   3.3 GB: 27-of-99   4.9 GB: 28-of-99   5.1 GB: 29-of-99   5.1 GB: 30-of-99   4.9 GB: 31-of-99   5.1 GB: 32-of-99   5.1 GB: 33-of-99   4.9 GB: 34-of-99   3.3 GB: 35-of-99   4.9 GB: 36-of-99   5.1 GB: 37-of-99   5.1 GB: 38-of-99   4.9 GB: 39-of-99   5.1 GB: 40-of-99
Quantization Typeq8
Model ArchitectureKimiK25ForConditionalGeneration
Licenseother

Best Alternatives to Kimi K2.6 Mlx DQ3 K M Q8

Best Alternatives
Context / RAM
Downloads
Likes
Kimi K2.7 Code Mlx DQ3 K M Q80K / 186 GB38655
Kimi K2.7 Code 4bit0K / 142.3 GB33825
Kimi K2.5 Mlx DQ3 K M Q80K / 186 GB4433
Note: green Score (e.g. "73.2") means that the model is better than mlx-community/Kimi-K2.6-mlx-DQ3_K_M-q8.