LLM EXPLORER 60,828 MODELS INDEXED

MiniMax M2.7 AWQ 4bit by cyankiwi

By cyankiwi · 61471 downloads
HF Repo ↗ Copy URL

MiniMax M2.7 AWQ 4bit is an open-source language model by cyankiwi. Features: 36.8b LLM, VRAM: 130.5GB, Context: 192K, License: other, Quantized, LLM Explorer Score: 0.46, ELO: 1404.

  4bit   Awq Base model:minimaxai/minimax-m... Base model:quantized:minimaxai...   Compressed-tensors   Conversational   Custom code   Endpoints compatible   Minimax m2   Quantized   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/cyankiwi/MiniMax-M2.7-AWQ-4bit 

MiniMax M2.7 AWQ 4bit Benchmarks

LMSys ELO 1404 vs 1272 (so35) 10.4%
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

MiniMax M2.7 AWQ 4bit Parameters and Internals

LLM NameMiniMax M2.7 AWQ 4bit
Repository 🤗https://huggingface.co/cyankiwi/MiniMax-M2.7-AWQ-4bit 
Base Model(s)  MiniMaxAI/MiniMax-M2.7   MiniMaxAI/MiniMax-M2.7
Model Size36.8b
Required VRAM130.5 GB
Updated2026-07-31
Maintainercyankiwi
Model Typeminimax_m2
Model Files  5.0 GB: 1-of-27   5.0 GB: 2-of-27   5.0 GB: 3-of-27   5.0 GB: 4-of-27   5.0 GB: 5-of-27   5.0 GB: 6-of-27   5.0 GB: 7-of-27   5.0 GB: 8-of-27   5.0 GB: 9-of-27   5.0 GB: 10-of-27   5.0 GB: 11-of-27   5.0 GB: 12-of-27   5.0 GB: 13-of-27   5.0 GB: 14-of-27   5.0 GB: 15-of-27   5.0 GB: 16-of-27   5.0 GB: 17-of-27   5.0 GB: 18-of-27   5.0 GB: 19-of-27   5.0 GB: 20-of-27   5.0 GB: 21-of-27   5.0 GB: 22-of-27   5.0 GB: 23-of-27   5.0 GB: 24-of-27   5.0 GB: 25-of-27   4.3 GB: 26-of-27   1.2 GB: 27-of-27
AWQ QuantizationYes
Quantization Typeawq|4bit
Model ArchitectureMiniMaxM2ForCausalLM
Licenseother
Context Length196608
Model Max Length196608
Transformers Version4.57.6
Tokenizer ClassGPT2Tokenizer
Vocabulary Size200064