What are the hardware requirements for Instella MoE 16B A3B Pretrain?

Instella MoE 16B A3B Pretrain requires approximately 31.7 GB of VRAM and supports a context window of 4K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Instella MoE 16B A3B Pretrain and how large is it?

Instella MoE 16B A3B Pretrain is developed by amd, a model with 15.9b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Instella MoE 16B A3B Pretrain?

Instella MoE 16B A3B Pretrain is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Instella MoE 16B A3B Pretrain by amd — VRAM 31.7GB, 4K context

Name: Instella MoE 16B A3B Pretrain
Author: amd

Instella MoE 16B A3B Pretrain is an open-source language model by amd. Features: 15.9b LLM, VRAM: 31.7GB, Context: 4K, License: other, MoE, LLM Explorer Score: 0.27.

Arxiv:2511.10628 Amd Conversational Custom code Deepseek v3 En Endpoints compatible Instella Moe Region:us Rocm Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/amd/Instella-MoE-16B-A3B-Pretrain

Instella MoE 16B A3B Pretrain Parameters and Internals

LLM Name	Instella MoE 16B A3B Pretrain
Repository 🤗	https://huggingface.co/amd/Instella-MoE-16B-A3B-Pretrain
Model Size	15.9b
Required VRAM	31.7 GB
Updated	2026-07-25
Maintainer	amd
Model Type	deepseek_v3
Model Files	10.0 GB: 1-of-4 10.0 GB: 2-of-4 10.0 GB: 3-of-4 1.7 GB: 4-of-4
Supported Languages	en
Model Architecture	InstellaMoEForCausalLM
License	other
Context Length	4096
Model Max Length	4096
Transformers Version	4.57.1
Tokenizer Class	LlamaTokenizerFast
Beginning of Sentence Token	<｜begin▁of▁sentence｜>
End of Sentence Token	<｜end▁of▁sentence｜>
Vocabulary Size	128896
Torch Data Type	bfloat16

Rank the Instella MoE 16B A3B Pretrain capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to Instella MoE 16B A3B Pretrain

Best Alternatives	Context / RAM	Downloads	Likes
Instella MoE 16B A3B Base	64K / 31.9 GB	918	6
Instella MoE 16B A3B Think	32K / 31.9 GB	580	44
Instella MoE 16B A3B SFT	32K / 31.9 GB	89	6
Instella MoE 16B A3B DPO	32K / 31.9 GB	56	3
Instella MoE 16B A3B Midtrain	4K / 63.5 GB	70	2

Note: green Score (e.g. "73.2") means that the model is better than amd/Instella-MoE-16B-A3B-Pretrain.