Kimi K2 Base by unsloth

 ยป  All LLMs  ยป  unsloth  ยป  Kimi K2 Base   URL Share it on

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: 0.24.

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 ๐Ÿค—: https://huggingface.co/unsloth/Kimi-K2-Base 

Kimi K2 Base Benchmarks

nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Kimi K2 Base (unsloth/Kimi-K2-Base)
๐ŸŒŸ Advertise your project ๐Ÿš€

Kimi K2 Base Parameters and Internals

LLM NameKimi K2 Base
Repository ๐Ÿค—https://huggingface.co/unsloth/Kimi-K2-Base 
Base Model(s)  Kimi K2 Base   moonshotai/Kimi-K2-Base
Model Size1026.5b
Required VRAM652.7 GB
Updated2026-02-26
Maintainerunsloth
Model Typedeepseek_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 ArchitectureDeepseekV3ForCausalLM
Licenseother
Context Length131072
Model Max Length131072
Transformers Version4.53.2
Vocabulary Size163840
Torch Data Typebfloat16

Best Alternatives to Kimi K2 Base

Best Alternatives
Context / RAM
Downloads
Likes
DeepSeek V3 Base160K / 171.8 GB23451684
DeepSeek R1 0528 FP4160K / 107 GB7299136
DeepSeek R1 FP4160K / 232 GB6856264
DeepSeek V3.0324 FP4160K / 242.8 GB998412
...V3.1 Base Int4 Mixed AutoRound160K / 220 GB1032
...Seek V3.1 Int4 Mixed AutoRound160K / 225 GB2410
DeepSeek V3.1 BF16160K / 358.8 GB711
DeepSeek V3.1160K / 176.1 GB443
DeepSeek TNG R1T2 Chimera160K / 180.4 GB376
DeepSeek V3.0324160K / 376 GB358
Note: green Score (e.g. "73.2") means that the model is better than unsloth/Kimi-K2-Base.

Rank the Kimi K2 Base Capabilities

๐Ÿ†˜ Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! ๐ŸŒŸ

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  

What open-source LLMs or SLMs are you in search of? 51634 in total.

Our Social Media →  
Original data from HuggingFace, OpenCompass and various public git repos.
Check out Ag3ntum โ€” our secure, self-hosted AI agent for server management.
Release v20241124