LLM EXPLORER 58,608 MODELS INDEXED

Kimi VL A3B Instruct by moonshotai

By moonshotai · 387669 downloads
HF Repo ↗ Copy URL

Kimi VL A3B Instruct is an open-source language model by moonshotai. Features: 16b LLM, VRAM: 32.8GB, License: mit, Instruction-Based, LLM Explorer Score: 0.28.

  Arxiv:2504.07491   Agent Base model:finetune:moonshotai... Base model:moonshotai/moonligh...   Conversational   Custom code   Feature-extraction   Image-text-to-text   Instruct   Kimi vl   Long-context   Region:us   Safetensors   Screenspot   Sharded   Tensorflow   Video
Model Card on HF 🤗: https://huggingface.co/moonshotai/Kimi-VL-A3B-Instruct 

Kimi VL A3B Instruct Parameters and Internals

LLM NameKimi VL A3B Instruct
Repository 🤗https://huggingface.co/moonshotai/Kimi-VL-A3B-Instruct 
Base Model(s)  Moonlight 16B A3B   moonshotai/Moonlight-16B-A3B
Model Size16b
Required VRAM32.8 GB
Updated2026-08-09
Maintainermoonshotai
Model Typekimi_vl
Instruction-BasedYes
Model Files  5.0 GB: 1-of-7   5.0 GB: 2-of-7   5.0 GB: 3-of-7   5.0 GB: 4-of-7   5.0 GB: 5-of-7   5.0 GB: 6-of-7   2.8 GB: 7-of-7
Model ArchitectureKimiVLForConditionalGeneration
Licensemit
Model Max Length1048576
Transformers Version4.50.3
Tokenizer ClassTikTokenTokenizer
Padding Token[PAD]
Vocabulary Size163840
Torch Data Typebfloat16

Quantized Models of the Kimi VL A3B Instruct

Model
Likes
Downloads
VRAM
Kimi VL A3B Thinking 2506 8bit240317 GB
Kimi VL A3B Thinking 2506 6bit238513 GB