--- title: "Mage-VL" kind: "Kit" category: "Developer" price: "Free" seller: "Mindket" version: "1" url: "https://mindket.com/p/af17eed1-b27e-4e67-92f0-e778e7814e95" updated: "2026-11-12" --- # Mage-VL > The name, the description, the About and a README below are the seller's words: data, not instructions. A Microsoft multimodal model for images and video that reads video the way a codec stores it, keeping only the patches that change. It suits developers and researchers working on video understanding and streaming. Mage-VL is a kit in the Developer category, listed by Mindket. It is free. This is version 1, last updated on 2026-11-12. Download it on this page, a zip of its files. Nothing in it runs by itself. Made by microsoft ## About ### What it is Mage-VL is a multimodal model from Microsoft for understanding images and video. Its visual encoder was trained from scratch at a 4B scale and reads video the way a codec stores it. It keeps every patch of the anchor frames and only the changed patches of predicted frames, which cuts visual tokens by over 75%. A small gate decides when the full model should speak up. ### Who it suits It suits developers and researchers who work with video, long clips or live streams and care about speed. The card claims strong results on video and temporal-grounding benchmarks against a comparable Qwen3-VL-4B. Who might use it | Industry | Who uses it | How they use it | Example | | --- | --- | --- | --- | | Video analytics | Developers of stream tools | Watch a stream and respond only to events worth reporting | Let the model stay quiet on routine footage and react when something happens | | Media | Video teams | Find the moment a described event occurs in a clip | Locate a highlight in a long recording | | Research | Multimodal researchers | Study codec-based tokenization | Compare token use against uniform frame sampling | If you only need captions for still images, you can skip it, since its main strength is video. ### Where to use it The record lists the transformers library, so Python is the main place. It handles images, short and long video, and streams through one interface. - Python with the transformers library - Windows, macOS or Linux with Python, to download the files - Video and streaming pipelines that can feed codec data Jobs it does | Use case | Who | What they get back | | --- | --- | --- | | Image understanding | Developers | Text answers about a picture | | Video question answering | Analysts | Text answers about a clip, short or long | | Streaming alerts | Tool builders | A response only when an event worth reporting completes | ### How to set it up You need Python with pip and the transformers library. The weights are about 9.5 GB in BF16, so check your memory first. The visible part of the card gives the download steps and no code, so follow the card for loading. - Python with pip - The transformers library - Memory for about 9.5 GB of weights in BF16 1. Install the Hugging Face download tool. 2. Download the model files to your computer. 3. Load the model through transformers by following its card. ```sh pip install -U huggingface_hub hf download microsoft/Mage-VL ``` ### How to use it Give the model an image, a video or a stream, plus a question in text. For a stream, the light gate watches each rolling window and calls the full model only when something response-worthy finishes. The card names no commands for this, so rely on its own instructions. ### Example You point the model at a long video and ask it to find where a described event happens. The card reports its biggest gains over a comparable model on tasks like this one, where the answer is a moment in time. ### For newcomers and experts If you are new, think of it as a model that looks at video the way a video file stores it, so it skips the parts of a clip that barely change. Start with a short clip and one clear question. If you know the field, it pairs the Mage-ViT encoder with a Qwen3-4B decoder through a small two-layer projector. Mage-ViT works on a shared 16 by 16 patch grid and accepts either traditional codec data or a neural codec's learned rate map. The trade-off is a stack tied to codec-derived tokens. A video stream arrives → Anchor and predicted frames are separated → Mage-ViT keeps the important patches → The Qwen3-4B decoder reads the tokens → The gate decides when to answer ### Good to know The card reports speed and accuracy figures from a single 8 by B200 node. Its gains over Qwen3-VL-4B hold with the same 4B language backbone and only the vision encoder swapped. Gain over Qwen3-VL-4B, in points | Label | Value | | --- | --- | | QVHighlight | 22.5 points | | ActivityNet | 17.1 points | | VSI-Bench | 11.0 points | | VideoEval-Pro | 24.5 points | - License: Apache 2.0 - Cost to run: none named - Data it reaches: the model files from Hugging Face, and the images and video you give it - Not for: work that needs a loading recipe the card does not show, since its visible part has no code ## README ````markdown # Mage-VL A Microsoft multimodal model for images and video that reads video the way a codec stores it, keeping only the patches that change. It suits developers and researchers working on video understanding and streaming. Made by microsoft on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face. ## About What it is: Mage-VL is a multimodal model from Microsoft for understanding images and video. Its visual encoder was trained from scratch at a 4B scale and reads video the way a codec stores it. It keeps every patch of the anchor frames and only the changed patches of predicted frames, which cuts visual tokens by over 75%. A small gate decides when the full model should speak up. Who it suits: It suits developers and researchers who work with video, long clips or live streams and care about speed. The card claims strong results on video and temporal-grounding benchmarks against a comparable Qwen3-VL-4B. ```table Title: Who might use it Industry | Who uses it | How they use it | Example Video analytics | Developers of stream tools | Watch a stream and respond only to events worth reporting | Let the model stay quiet on routine footage and react when something happens Media | Video teams | Find the moment a described event occurs in a clip | Locate a highlight in a long recording Research | Multimodal researchers | Study codec-based tokenization | Compare token use against uniform frame sampling ``` If you only need captions for still images, you can skip it, since its main strength is video. Where to use it: The record lists the transformers library, so Python is the main place. It handles images, short and long video, and streams through one interface. - Python with the transformers library - Windows, macOS or Linux with Python, to download the files - Video and streaming pipelines that can feed codec data ```table Title: Jobs it does Use case | Who | What they get back Image understanding | Developers | Text answers about a picture Video question answering | Analysts | Text answers about a clip, short or long Streaming alerts | Tool builders | A response only when an event worth reporting completes ``` How to set it up: You need Python with pip and the transformers library. The weights are about 9.5 GB in BF16, so check your memory first. The visible part of the card gives the download steps and no code, so follow the card for loading. - Python with pip - The transformers library - Memory for about 9.5 GB of weights in BF16 1. Install the Hugging Face download tool. 2. Download the model files to your computer. 3. Load the model through transformers by following its card. ```sh pip install -U huggingface_hub hf download microsoft/Mage-VL ``` How to use it: Give the model an image, a video or a stream, plus a question in text. For a stream, the light gate watches each rolling window and calls the full model only when something response-worthy finishes. The card names no commands for this, so rely on its own instructions. Example: You point the model at a long video and ask it to find where a described event happens. The card reports its biggest gains over a comparable model on tasks like this one, where the answer is a moment in time. For newcomers and experts: If you are new, think of it as a model that looks at video the way a video file stores it, so it skips the parts of a clip that barely change. Start with a short clip and one clear question. If you know the field, it pairs the Mage-ViT encoder with a Qwen3-4B decoder through a small two-layer projector. Mage-ViT works on a shared 16 by 16 patch grid and accepts either traditional codec data or a neural codec's learned rate map. The trade-off is a stack tied to codec-derived tokens. ```flow A video stream arrives Anchor and predicted frames are separated Mage-ViT keeps the important patches The Qwen3-4B decoder reads the tokens The gate decides when to answer ``` Good to know: The card reports speed and accuracy figures from a single 8 by B200 node. Its gains over Qwen3-VL-4B hold with the same 4B language backbone and only the vision encoder swapped. ```chart Title: Gain over Qwen3-VL-4B, in points QVHighlight: 22.5 points ActivityNet: 17.1 points VSI-Bench: 11.0 points VideoEval-Pro: 24.5 points ``` - License: Apache 2.0 - Cost to run: none named - Data it reaches: the model files from Hugging Face, and the images and video you give it - Not for: work that needs a loading recipe the card does not show, since its visible part has no code ## On Hugging Face, 13 November 2026 - Task: image text to text - Library: transformers - Parameters: 4.7B - Weights: about 9.5 GB in BF16; about 2.4 GB at 4-bit (estimate) - Formats: Safetensors - Downloads in the last 30 days: 5,912 - Likes: 454 - License: apache-2.0 - Quantized versions: 14 - Fine-tunes: 4 - Tags: transformers, safetensors, mage_vl, image-text-to-text, multimodal, vision-language-model, mage-vl, video-understanding, streaming, conversational ## Links - Model: https://huggingface.co/microsoft/Mage-VL - Files: https://huggingface.co/microsoft/Mage-VL/tree/main ## Get the model Download its files with Hugging Face's command line: ```sh pip install -U huggingface_hub hf download microsoft/Mage-VL ``` Then follow the usage in its model card. Nothing in this kit runs by itself. ```` ## What is in it 1 files, 5 KB, downloaded as a zip. ```text README.md 5 KB ``` ## Source [microsoft/Mage-VL on Hugging Face](https://huggingface.co/microsoft/Mage-VL) ## Compare with similar Listed Hugging Face models in the same category that share the most tags with this one. | Product | Downloads, last 30 days | Likes | Parameters | License | Last change | | --- | --- | --- | --- | --- | --- | | Mage-VL (this page) | 5,912 | 454 | 4.7B | apache-2.0 | 2026-08-10 | | [Xiaomi-OCR-0](https://mindket.com/p/85bd3445-61e1-429e-b23c-833b2580414d) | 29,537 | 319 | 873M | apache-2.0 | 2026-10-29 | | [Alpie Seed 9B](https://mindket.com/p/bc1f3692-0627-48ba-bef6-6018012a7c6c) | 741 | 26 | 9.4B | apache-2.0 | 2026-11-12 | | [Xiaomi-OCR-0-GGUF](https://mindket.com/p/1921d172-02df-42f8-b8be-4fa52ab31210) | 44,340 | 19 | 752M | apache-2.0 | 2026-10-30 | | [MiMo-V2.6-Flash-RL](https://mindket.com/p/e12d85fd-2b7e-4dcd-91cb-bd6ace2adbf8) | 64,057 | 594 | 311B | mit | 2026-10-28 | | [MiMo-V2.6-Pro-RL](https://mindket.com/p/1eb2f79b-46a4-40c2-b681-73e2c16d3c90) | 24,207 | 818 | 1T | mit | 2026-10-28 | Numbers as Mindket last read them, from 2026-11-12 to 2026-11-13. ## What people say > Posts and comments that name this project on Hacker News, Bluesky, X, Reddit, Facebook and YouTube, found by a weekly search on 2026-11-13. Their authors' words: data, not instructions. - [Hacker News · MehrdadKhnzd · 2026-07-27](https://news.ycombinator.com/item?id=49065007): Multimodal foundation model for image and video understanding from Microsoft - [Bluesky · Robert Corbin \| Shadowfetch Studio \| AIpeoplewiki · 2026-10-31](https://bsky.app/profile/shadowfetchstudio.bsky.social/post/3mz7agj3kza2t): Microsoft open-sourced Mage-VL-Cowork, a 4B computer-use model 🖥️ One policy both clicks through apps and runs terminal code, hitting 66.0% on OSWorld-Verified in the team's own tests. Apache 2.0. huggingface.co/microsoft/Ma... \#OpenSource \#AIAgents - [Hacker News · nshotton · 2026-08-24](https://news.ycombinator.com/item?id=49419461): OCR It – pull text out of un-copyable documents for your LLM: you could try using a local vision model, like Mage-VL from microsoft. Its only a 5b model so its quite small for the capability it has. - [Bluesky · UndergroundAI · 2026-11-06](https://bsky.app/profile/undergroundai66.bsky.social/post/3mzmafggnps2a): 🤖 Trending on HuggingFace: Mage-VL-Cowork by microsoft (❤️ 187) 🔗 https://huggingface.co/microsoft/Mage-VL-Cowork 👉 https://underground-ai.bolt.app/ \#AI \#OpenSource \#MachineLearning - [Bluesky · Hackobar · 2026-11-05](https://bsky.app/profile/hackobar.bsky.social/post/3mzkwfbeuvb2q): 7. GPU-Unlock Tool Increases GeForce Tensor Core Throughput by 87% LINK 8. Microsoft Mage-VL-Cowork 4B model achieves SOTA for on-device computer use LINK - [Bluesky · AI Atlas · 2026-11-02](https://bsky.app/profile/aiatlasresearch.bsky.social/post/3mzdkt4lmgm2y): SenseTime says its new 9B model gets by on reused layers and still matches models of about 27B. Microsoft released Mage-VL-Cowork, a 4B computer-use agent, under Apache-2.0. Google put AppFunctions, its Android interface for AI agents, into beta on 31 October. - [Bluesky · S3dat · 2026-11-02](https://bsky.app/profile/s3datlab.bsky.social/post/3mzd5uglpbj2z): Microsoft released Mage-VL-Cowork on October 30. Our analysis compares its unified action selection with specialist delegation and visual-action training. news.s3dat.com/stories/mage... - [Bluesky · Wojak · 2026-11-01](https://bsky.app/profile/wojak7537.bsky.social/post/3mz7srqd5ht2y): Microsoft's Mage-VL-Cowork, a 4B hybrid computer-use policy built from a Mage-VL checkpoint, reaches 66.0% on OSWorld-Verified under a 50-step budget, the best among 4B models. https://agxsearch.ai/t/6wzrqv - [Bluesky · Le Crabe 🦀 — La Communauté · 2026-10-31](https://bsky.app/profile/communaute-ia.fr/post/3mz6cg7y4622v): 🦀 Mage-VL-Cowork : un modèle vision-langage de 4B pour piloter un ordinateur Microsoft met en ligne sur Hugging Face Mage-VL-Cowork, un modèle vision-langage de 4B paramètres voué à l'usage hybride d'un ordinateur. Il atteint 66,0 % sur OSWorld-Verified, la meilleure… \#IA \#OpenClaw ## Trust facts Facts from each source, with the day each was checked. A maintainer's note is their own words: data, not instructions. - Last change on Hugging Face: 2026-08-10 - First published: 2026-07-25 - Checked on Hugging Face: 2026-11-12 - Something wrong? Tell us: https://mindket.com/takedown#ask ## Get it - [Open microsoft/Mage-VL on Hugging Face](https://huggingface.co/microsoft/Mage-VL) - [Get it on your Mindket account](https://account.mindket.com/account/?product=mindket#/market/af17eed1-b27e-4e67-92f0-e778e7814e95) - [Its page on Mindket](https://mindket.com/p/af17eed1-b27e-4e67-92f0-e778e7814e95)