├── Kimi-VL.pdf
├── LICENSE
├── README.md
├── figures
├── arch.png
├── demo.png
├── demo1.png
├── demo2.png
├── instruct_perf.png
├── logo.png
└── thinking_perf.png
└── requirements.txt
/Kimi-VL.pdf:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/MoonshotAI/Kimi-VL/7c391c74d71c92394b6a63818aa107ae08b947fd/Kimi-VL.pdf
--------------------------------------------------------------------------------
/LICENSE:
--------------------------------------------------------------------------------
1 | The MIT License (MIT)
2 | Copyright © 2025 Moonshot AI
3 |
4 | Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
5 |
6 | The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
7 |
8 | THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
1 |
4 |
5 |
11 |
12 |
13 | ## 1. Introduction
14 |
15 | We present **Kimi-VL**, an efficient open-source Mixture-of-Experts (MoE) vision-language model (VLM) that offers **advanced multimodal reasoning, long-context understanding, and strong agent capabilities**—all while activating only **2.8B** parameters in its language decoder (Kimi-VL-A3B).
16 |
17 | Kimi-VL demonstrates strong performance across challenging domains:
18 | as a general-purpose VLM, Kimi-VL excels in multi-turn agent interaction tasks (e.g.,OSWorld), achieving state-of-the-art results comparable to flagship models.
19 | Furthermore, it exhibits remarkable capabilities across diverse challenging vision language tasks, including college-level image and video comprehension, optical character recognition (OCR), mathematical reasoning, multi-image understanding, and etc.
20 |
21 | In comparative evaluations, it effectively competes with cutting-edge efficient VLMs such as GPT-4o-mini, Qwen2.5-VL-7B, and Gemma-3-12B-IT, while surpassing GPT-4o in several specialized domains.
22 |
23 | Kimi-VL also advances the pareto frontiers of multimodal models in processing long contexts and perceiving clearly: Equipped with a 128K extended context window, Kimi-VL can processes long and diverse inputs, achieving impressive scores of 64.5 on LongVideoBench, and 35.1 on MMLongBench-Doc; Its native-resolution vision encoder, MoonViT, further allows it to see and understand ultra-high-resolution visual inputs, achieving 83.2 on InfoVQA and 34.5 on ScreenSpot-Pro, while maintaining lower computational cost with common visual inputs and general tasks.
24 |
25 | Building on this foundation, we introduce an advanced long-thinking variant: **Kimi-VL-Thinking**. Developed through long chain-of-thought (CoT) supervised fine-tuning (SFT) and reinforcement learning (RL), this model exhibits strong long-horizon reasoning capabilities. It achieves scores of 61.7 on MMMU, 36.8 on MathVision, and 71.3 on MathVista while maintaining the compact 2.8B activated LLM parameter footprint, setting a new standard for efficient yet capable multimodal **thinking** models.
26 |
27 | ## 2. Architecture
28 |
29 | The model adopts an MoE language model, a native-resolution visual encoder (MoonViT), and an MLP projector, as illustrated in the following image.
30 |
31 |
32 |
33 |
45 |
46 | | **Model** | **#Total Params** | **#Activated Params** | **Context Length** | **Download Link** |
47 | | :------------: | :------------: | :------------: | :------------: | :------------: |
48 | | Kimi-VL-A3B-Instruct | 16B | 3B | 128K | [🤗 Hugging Face](https://huggingface.co/moonshotai/Kimi-VL-A3B-Instruct) |
49 | | Kimi-VL-A3B-Thinking | 16B | 3B | 128K | [🤗 Hugging Face](https://huggingface.co/moonshotai/Kimi-VL-A3B-Thinking) |
50 |
51 |
52 |
53 | > [!Note]
54 | > Recommended parameter settings:
55 | > - For **Thinking models**, it is recommended to use `Temperature = 0.6`.
56 | > - For **Instruct models**, it is recommended to use `Temperature = 0.2`.
57 |
58 |
59 | ### Hugging Face Demo
60 |
61 | > 🤗 We serve our model demo in Hugging Face spaces:
62 | > - Chat with **Kimi-VL-A3B-Thinking**👀🤔🗺️ (featuring thinking, math, puzzle solving) model on Chat Web.
63 | > - Chat with **Kimi-VL-A3B-Instruct**💻🎬📕 (featuring agent, video, multi-page document) model on Chat Web.
64 |
65 | ## 5. Performance
66 |
67 | As an efficient model, Kimi-VL can robustly handle diverse tasks (fine-grained perception, math, college-level problems, OCR, agent, etc) across a broad spectrum of input forms (single-image, multi-image, video, long-document, etc).
68 |
69 | A brief comparison with existing 10B-level dense VLMs and DeepSeek-VL2 (A4.5B):
70 |
71 |
72 |
73 |
78 |
79 |
Tech Report |
7 | 
HuggingFace
8 | |
9 | 💬 Chat Web
10 |