What are the hardware requirements for Kimi K2 Base?

Kimi K2 Base requires approximately 652.7 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 Kimi K2 Base and how large is it?

Kimi K2 Base is developed by unsloth, a model with 1026.5b 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 Kimi K2 Base?

Kimi K2 Base is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Kimi K2 Base by unsloth — VRAM 652.7GB, 128K context

Name: Kimi K2 Base
Author: unsloth

Kimi K2 Base is an open-source language model by unsloth. Features: 1026.5b LLM, VRAM: 652.7GB, Context: 128K, License: other, LLM Explorer Score: 0.19.

Base model:moonshotai/kimi-k2-... Base model:quantized:moonshota... Conversational Custom code Deepseek v3 Endpoints compatible Fp8 Region:us Safetensors Sharded Tensorflow Unsloth

Model Card on HF 🤗: https://huggingface.co/unsloth/Kimi-K2-Base

Kimi K2 Base Parameters and Internals

LLM Name	Kimi K2 Base
Repository 🤗	https://huggingface.co/unsloth/Kimi-K2-Base
Base Model(s)	Kimi K2 Base moonshotai/Kimi-K2-Base
Model Size	1026.5b
Required VRAM	652.7 GB
Updated	2026-06-04
Maintainer	unsloth
Model Type	deepseek_v3
Model Files	2.9 GB: 1-of-61 17.1 GB: 10-of-61 17.1 GB: 11-of-61 17.1 GB: 12-of-61 17.1 GB: 13-of-61 17.1 GB: 14-of-61 17.1 GB: 15-of-61 17.1 GB: 16-of-61 17.1 GB: 17-of-61 17.1 GB: 18-of-61 17.1 GB: 19-of-61 17.1 GB: 2-of-61 17.1 GB: 20-of-61 17.1 GB: 21-of-61 17.1 GB: 22-of-61 17.1 GB: 23-of-61 17.1 GB: 24-of-61 17.1 GB: 25-of-61 17.1 GB: 26-of-61 17.1 GB: 27-of-61 17.1 GB: 28-of-61 17.1 GB: 29-of-61 17.1 GB: 3-of-61 17.1 GB: 30-of-61 17.1 GB: 31-of-61 17.1 GB: 32-of-61 17.1 GB: 33-of-61 17.1 GB: 34-of-61 17.1 GB: 35-of-61 17.1 GB: 36-of-61 17.1 GB: 37-of-61 17.1 GB: 38-of-61 17.1 GB: 39-of-61 17.1 GB: 4-of-61 17.1 GB: 40-of-61 17.1 GB: 41-of-61 17.1 GB: 42-of-61 17.1 GB: 43-of-61 17.1 GB: 44-of-61
Model Architecture	DeepseekV3ForCausalLM
License	other
Context Length	131072
Model Max Length	131072
Transformers Version	4.53.2
Vocabulary Size	163840
Torch Data Type	bfloat16

Rank the Kimi K2 Base 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 Kimi K2 Base

Best Alternatives	Context / RAM	Downloads	Likes
Kimi K2 Thinking NVFP4	256K / 210 GB	15569	29
Kimi K2 Thinking BF16	256K / 220 GB	63	6
Kimi K2 Instruct 0905	256K / 667.4 GB	17	7
DeepSeek R1 0528	160K / 176.1 GB	48	15
DeepSeek R1 0528 FP4	160K / 107 GB	72991	36
DeepSeek V3.0324 FP4	160K / 242.8 GB	9984	12
DeepSeek R1 FP4	160K / 232 GB	6856	264
DeepSeek V3.1 MLX 5.5bit	160K / GB	400	5
...V3.1 Base Int4 Mixed AutoRound	160K / 220 GB	103	2
...Seek V3.1 Int4 Mixed AutoRound	160K / 225 GB	53	11

Note: green Score (e.g. "73.2") means that the model is better than unsloth/Kimi-K2-Base.