LLMs · Microsoft

Mage-VL-Cowork 4.7B New

Name: Mage-VL-Cowork 4.7B
Author: Microsoft

Parameters: 4.7B
Architecture: Dense
Max context: 32K tokens
License: apache-2.0
Released: 2026-10

How to run it

Original weights · Hugging Face

Full-precision files from Microsoft, for fine-tuning, vLLM, Transformers.

huggingface.co/microsoft/Mage-VL-Cowork (opens in a new tab)

No sign-in needed.

Official repo · GitHub

Reference code, inference scripts and docs from Microsoft. The weights themselves download from Hugging Face.

github.com/microsoft/Mage (opens in a new tab)

Microsoft’s Mage repo; the mage_vl_cowork folder has the agent harness and evaluation code.

git clone https://github.com/microsoft/Mage

Links checked 5 days ago against Hugging Face, Ollama and GitHub.

Apps that run Mage-VL-Cowork 4.7B

Our pick for each kind of computer, assuming the model fits in its memory. Pick your own GPU or device in the Model Finder for an exact answer.

Mac (Apple silicon)

Docker Model Runner

Handy if you already work in Docker.

Also good:

Text Generation WebUI

Windows PC, NVIDIA GPU

Docker Model Runner

Handy if you already work in Docker.

Also good:

Windows PC, AMD Radeon

Docker Model Runner

Handy if you already work in Docker.

Also good:

Text Generation WebUI

Linux, NVIDIA GPU

Docker Model Runner

Handy if you already work in Docker.

Also good:

Linux server, 80 GB+ GPUs

vLLM

The full model fits in GPU memory at 8-bit, so it can batch many requests: best for serving a team or an app.

Also good:

App	Type	Opens it as	Systems	Engine	Start
vLLM	Server	Original weights	WSLWindowsLinux	vLLM	`vllm serve microsoft/Mage-VL-Cowork`
SGLang	Server	Original weights	WSLWindowsLinux	SGLang	`python3 -m sglang.launch_server --model-path microsoft/Mage-VL-Cowork`
Docker Model Runner	CLI + API	Original weights	macOSWindowsLinux	llama.cpp, vLLM	`docker model run hf.co/microsoft/Mage-VL-Cowork`
Text Generation WebUI	Desktop app	Original weights	macOSWindowsLinux	llama.cpp, Transformers, ExLlamaV3	Get Text Generation WebUI

Which apps can open a model depends on the files published for it: GGUF for llama.cpp-based apps, MLX for Apple’s engine, the original weights for vLLM and SGLang. Compare every app.

Memory needed

Weights plus KV cache and runtime overhead. Leave a little extra free for the operating system and other apps. Try another context length.

Quantization	Quality	Weights	Total at 8K context	Total at 32K context	Download
`Q2_K`	Low	1.9 GB	3.7 GB	7.3 GB	All files (opens in a new tab)
`Q3_K_M`	Low	2.3 GB	4.1 GB	7.8 GB	All files (opens in a new tab)
`Q4_K_M`	Good	2.9 GB	4.7 GB	8.3 GB	All files (opens in a new tab)
`Q5_K_M`	High	3.4 GB	5.2 GB	8.8 GB	All files (opens in a new tab)
`Q6_K`	High	3.9 GB	5.7 GB	9.3 GB	All files (opens in a new tab)
`Q8_0`	Very high	5.0 GB	6.8 GB	10.5 GB	All files (opens in a new tab)
`F16`	Full	9.5 GB	11.3 GB	14.9 GB	Original weights (opens in a new tab)

Best hardware for Mage-VL-Cowork 4.7B

The least expensive GPU in our catalog that runs it without offloading is the Intel Arc A380 (launch price $139) at Q4_K_M.

Quality	Hardware	Quant	Memory	Speed	Runs
60Good quality (B) · 60 of 100: how capable the model is at F16, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 4070 Ti SUPER	`F16`	12.1 GB / 16.0 GB	40.8 tok/s	Runs great
60Good quality (B) · 60 of 100: how capable the model is at F16, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 4080 SUPER	`F16`	12.1 GB / 16.0 GB	44.7 tok/s	Runs great
60Good quality (B) · 60 of 100: how capable the model is at F16, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 5070 Ti	`F16`	12.1 GB / 16.0 GB	54.4 tok/s	Runs great
60Good quality (B) · 60 of 100: how capable the model is at F16, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 5080	`F16`	12.1 GB / 16.0 GB	58.3 tok/s	Runs great
59Good quality (B) · 59 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 4060	`Q6_K`	6.1 GB / 8.0 GB	48.9 tok/s	Runs great
60Good quality (B) · 60 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking.	Intel Arc B580	`Q8_0`	7.4 GB / 12.0 GB	45.9 tok/s	Runs great
60Good quality (B) · 60 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 3060 12GB	`Q8_0`	7.4 GB / 12.0 GB	47.5 tok/s	Runs great
60Good quality (B) · 60 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 4070 SUPER	`Q8_0`	7.4 GB / 12.0 GB	66.4 tok/s	Runs great
60Good quality (B) · 60 of 100: how capable the model is at F16, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking.	AMD Radeon RX 7900 XTX	`F16`	12.5 GB / 24.0 GB	50.2 tok/s	Runs great
60Good quality (B) · 60 of 100: how capable the model is at F16, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking.	NVIDIA GeForce RTX 3090	`F16`	12.5 GB / 24.0 GB	56.9 tok/s	Runs great

Check it on your own hardware

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.