What are the hardware requirements for K2 Merged 3.5T Bf16?

K2 Merged 3.5T Bf16 requires approximately 779.9 GB of VRAM and supports a context window of 128K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed K2 Merged 3.5T Bf16 and how large is it?

K2 Merged 3.5T Bf16 is developed by NousResearch, a model with 3468.1b 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 K2 Merged 3.5T Bf16?

K2 Merged 3.5T Bf16 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

K2 Merged 3.5T Bf16 by NousResearch — VRAM 779.9GB, 128K context

Name: K2 Merged 3.5T Bf16
Author: NousResearch

K2 Merged 3.5T Bf16 is an open-source language model by NousResearch. Features: 3468.1b LLM, VRAM: 779.9GB, Context: 128K, License: other, Merged, LLM Explorer Score: 0.21.

Merged Model Conversational Custom code Deepseek v3 Endpoints compatible Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/NousResearch/k2-merged-3.5T-bf16

K2 Merged 3.5T Bf16 Parameters and Internals

LLM Name	K2 Merged 3.5T Bf16
Repository 🤗	https://huggingface.co/NousResearch/k2-merged-3.5T-bf16
Merged Model	Yes
Model Size	3468.1b
Required VRAM	779.9 GB
Updated	2026-08-06
Maintainer	NousResearch
Model Type	deepseek_v3
Model Files	20.0 GB: 1-of-348 20.0 GB: 2-of-348 20.0 GB: 3-of-348 20.0 GB: 4-of-348 20.0 GB: 5-of-348 20.0 GB: 6-of-348 20.0 GB: 7-of-348 20.0 GB: 8-of-348 20.0 GB: 9-of-348 19.9 GB: 10-of-348 20.0 GB: 11-of-348 20.0 GB: 12-of-348 20.0 GB: 13-of-348 20.0 GB: 14-of-348 20.0 GB: 15-of-348 20.0 GB: 16-of-348 20.0 GB: 17-of-348 20.0 GB: 18-of-348 20.0 GB: 19-of-348 20.0 GB: 20-of-348 20.0 GB: 21-of-348 20.0 GB: 22-of-348 20.0 GB: 23-of-348 20.0 GB: 24-of-348 20.0 GB: 25-of-348 20.0 GB: 26-of-348 20.0 GB: 27-of-348 20.0 GB: 28-of-348 20.0 GB: 29-of-348 20.0 GB: 30-of-348 20.0 GB: 31-of-348 20.0 GB: 32-of-348 20.0 GB: 33-of-348 20.0 GB: 34-of-348 20.0 GB: 35-of-348 20.0 GB: 36-of-348 20.0 GB: 37-of-348 20.0 GB: 38-of-348 20.0 GB: 39-of-348
Model Architecture	DeepseekV3ForCausalLM
License	other
Context Length	131072
Model Max Length	131072
Transformers Version	4.55.4
Vocabulary Size	163840
Torch Data Type	bfloat16

Rank the K2 Merged 3.5T Bf16 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