LLM EXPLORER 67,808 MODELS INDEXED

Instella MoE 16B A3B Think SOMPOA Heresy by MuXodious

By MuXodious · 248 downloads

Instella MoE 16B A3B Think SOMPOA Heresy is an open-source language model by MuXodious. Features: 15.9b LLM, VRAM: 31.7GB, Context: 32K, License: other, MoE, LLM Explorer Score: 0.32.

  Arxiv:2609.00791   Abliterated   Amd Base model:amd/instella-moe-16... Base model:finetune:amd/instel...   Conversational   Custom code   Decensored   Deepseek v3   En   Endpoints compatible   Heretic   Instella   Moe   Region:us   Rocm   Safetensors   Sharded   Tensorflow   Uncensored

Instella MoE 16B A3B Think SOMPOA Heresy Parameters and Internals

LLM NameInstella MoE 16B A3B Think SOMPOA Heresy
Repository 🤗https://huggingface.co/MuXodious/Instella-MoE-16B-A3B-Think-SOMPOA-heresy 
Base Model(s)  Instella MoE 16B A3B Think   amd/Instella-MoE-16B-A3B-Think
Model Size15.9b
Required VRAM31.7 GB
Updated2026-11-18
MaintainerMuXodious
Model Typedeepseek_v3
Model Files  4.7 GB: 1-of-7   4.7 GB: 2-of-7   4.7 GB: 3-of-7   4.7 GB: 4-of-7   4.7 GB: 5-of-7   4.7 GB: 6-of-7   3.5 GB: 7-of-7
Supported Languagesen
Model ArchitectureInstellaMoEForCausalLM
Licenseother
Context Length32768
Model Max Length32768
Transformers Version4.57.1
Tokenizer ClassPreTrainedTokenizerFast
Beginning of Sentence Token<|begin▁of▁sentence|>
End of Sentence Token<|end▁of▁sentence|>
Vocabulary Size128896
Torch Data Typebfloat16

Best Alternatives to Instella MoE 16B A3B Think SOMPOA Heresy

Best Alternatives
Context / RAM
Downloads
Likes
Instella MoE 16B A3B Base64K / 31.9 GB9186
Instella MoE 16B A3B Think32K / 31.9 GB58044
Instella MoE 16B A3B SFT32K / 31.9 GB896
Instella MoE 16B A3B DPO32K / 31.9 GB563
Instella MoE 16B A3B Pretrain4K / 31.7 GB702
Instella MoE 16B A3B Midtrain4K / 63.5 GB702