AlibabaQwen 3Qwen3VLForConditionalGeneration

Qwen3 VL 32B Instruct — Hardware Requirements & GPU Compatibility

VisionFunctions

Qwen3 VL 32B Instruct is Alibaba's largest dense model in the initial Qwen3-VL lineup, a 33.4-billion-parameter vision-language model built to process images and text in one pass. It handles image description, visual question answering, and document understanding, and its visual-agent tuning lets it read GUI screenshots and reason about on-screen elements. Local inference calls for quantization and a capable GPU, fitting on a single high-end consumer or workstation card. The model supports a 262,144 token context window, enough for long documents or extended multi-turn conversations. It is released under the Apache 2.0 license, permitting unrestricted commercial and research use. Published in October 2025 alongside the 2B and 8B models, it shares the family's long native context and video-understanding capabilities, giving stronger multimodal reasoning than the smaller variants.

33.4B params · dense262K contextApache 2.0Oct 2025165.4K downloads257 likes377.7K quant downloads

Your hardware

Verdicts below are for the reference RTX 4090. Set yours to personalise the page.

Detection reads your browser's GPU name with no data sent anywhere.

Decent on a 24 GB GPU

Runs from 16 GB (IQ3_XXS · 16K)Best on 24 GB (Q4_K_M · ~32 tok/s · 16K)Macs from 16 GB (IQ2_XXS · ~14 tok/s)

Apache 2.0

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Qwen3 VL 32B Instruct

One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.

Ways to runOther ways

each opens step-by-step

Run in cloud

Your own private copy · Q4_K_M · 32 GB GPU

The easiest way to try this model
Running in a few minutes, easy setup
Chat right here or connect any app

$5+ bonus for new users

Run Qwen3 VL 32B Instruct

23.0 GB VRAM at 16K context · fits the RTX 4090 (1.0 GB free)

Works on macOS, Windows and Linux.

Install Ollama (Linux)
$curl -fsSL https://ollama.com/install.sh | sh
Install Ollama (macOS / Windows)
Download and run the app; it installs the ollama command.
Download Ollama ↗
Run Q4_K_M
$ollama run qwen3-vl:32b-instruct
Official Ollama library build. For the exact Q4_K_M file from Qwen/Qwen3-VL-32B-Instruct-GGUF: ollama run hf.co/Qwen/Qwen3-VL-32B-Instruct-GGUF:Q4_K_M
Raise the context window
Type it inside the chat. For the API server instead, start it with OLLAMA_CONTEXT_LENGTH=16384 ollama serve.
$/set parameter num_ctx 16384
A larger context needs more memory than the weights alone.
Use the API
Ollama serves its API on http://localhost:11434.

Works on macOS, Windows and Linux.

Install LM Studio
Desktop app for macOS (Apple silicon), Windows and Linux.
lmstudio.ai ↗
Download the model
In the app press Ctrl/Cmd + Shift + M and search qwen/qwen3-vl-32b, then pick Q4_K_M. Or use the CLI (run LM Studio once first):
$lms get qwen/qwen3-vl-32b@q4_k_m
Open on Hugging Face ↗
Load and chat
Open the model loader, select the model and set Context Length to 16384. Leave GPU offload at maximum if it fits.
Start the local server (optional)
$lms server start

Works on macOS, Windows and Linux.

Install llama.cpp (macOS)
MacPorts, Nix and conda-forge also work.
$brew install llama.cpp
Install options ↗
Install llama.cpp (Windows)
$winget install llama.cpp
Install llama.cpp (Linux)
Download the prebuilt llama-*-bin-ubuntu-* archive for your GPU backend (CPU, Vulkan) from the releases page, unpack it and run the binaries from that folder. Homebrew (brew install llama.cpp) and conda-forge also work on Linux.
llama.cpp releases ↗
Start the server
Downloads the model on first run. The chat UI and an OpenAI-compatible API are at http://localhost:8080.
$llama-server -hf Qwen/Qwen3-VL-32B-Instruct-GGUF:Q4_K_M -c 16384 -ngl 99 --port 8080

Works on macOS, Windows and Linux.

Option A: Docker Model Runner
Docker Desktop: Settings > AI > Enable Docker Model Runner. Docker Engine on Linux: install the docker-model-plugin package.
$docker model run hf.co/Qwen/Qwen3-VL-32B-Instruct-GGUF:Q4_K_M
Model Runner docs ↗
The quant tag after the repo is how Hugging Face cards write it; if your Docker version rejects it, drop the tag and pick the quant by repo.
Option B: llama.cpp server image (NVIDIA)
Needs the NVIDIA Container Toolkit. Models are cached in the llama-cache volume. Open http://localhost:8080.
$docker run --gpus all -p 8080:8080 -v llama-cache:/root/.cache ghcr.io/ggml-org/llama.cpp:server-cuda -hf Qwen/Qwen3-VL-32B-Instruct-GGUF:Q4_K_M -c 16384 -ngl 99 --host 0.0.0.0 --port 8080
Image tags (cuda13, vulkan, rocm) ↗
Cache path inside the image is not in the docs; remove the -v flag if the volume complains.

Linux only.

Install vLLM
Linux only; on Windows run it inside WSL2. NVIDIA GPU shown; AMD ROCm, Intel and CPU builds have their own install pages.
$uv venv --python 3.12 --seed
$source .venv/bin/activate
$uv pip install vllm --torch-backend=auto
vLLM installation ↗
Serve the model
OpenAI-compatible API at http://localhost:8000/v1.
$vllm serve Qwen/Qwen3-VL-32B-Instruct --max-model-len 16384
Runs the original weights, not the Q4_K_M GGUF; needs ~73 GB of GPU memory.

macOS only.

MLX supports only macOS, so follow these steps for that setup.

Install mlx-lm
$pip install mlx-lm
Chat in the terminal
$mlx_lm.chat --model lmstudio-community/Qwen3-VL-32B-Instruct-MLX-4bit
Downloads the model on first run. The MLX repo has its own quantisation; it may not match the quant selected above.
Or start a local server
Listens on http://localhost:8080.
$mlx_lm.server --model lmstudio-community/Qwen3-VL-32B-Instruct-MLX-4bit

Works on macOS, Windows and Linux.

Install Jan
jan.ai ↗
Find the model
Open the Hub tab, paste this model ID in Search, and choose Q4_K_M if offered.
$Qwen/Qwen3-VL-32B-Instruct-GGUF
Use the local API (optional)
Settings > Local API Server > Start Server. It listens on http://127.0.0.1:1337.

Rent GPUs hourly to run Q4_K_M right there. Pricing comes from partner marketplaces and varies.

Available GPU rentals for Q4_K_M, prioritized by value
GPU	Speed	$/1M tokens	Rent
RTX 5090 (32 GB)CheapestCheapest9 GB free	~56 tok/s	$2.45	Deploy on Vast.aifrom $0.50/hr (opens in new tab)	32 GB · 9 GB free · ~56 tok/s · $2.45 / 1M tokens
RTX PRO 4500 (32 GB)9 GB free	~28 tok/s	$3.35	Deploy on RunPodfrom $0.34/hr (opens in new tab)	32 GB · 9 GB free · ~28 tok/s · $3.35 / 1M tokens
A40 (48 GB)25 GB free	~22 tok/s	$4.44	Deploy on RunPodfrom $0.35/hr (opens in new tab)	48 GB · 25 GB free · ~22 tok/s · $4.44 / 1M tokens
RTX PRO 4500 (32 GB)9 GB free	~28 tok/s	$3.50	Deploy on Vast.aifrom $0.36/hr (opens in new tab)	32 GB · 9 GB free · ~28 tok/s · $3.50 / 1M tokens
RTX A6000 (48 GB)25 GB free	~24 tok/s	$4.64	Deploy on Vast.aifrom $0.40/hr (opens in new tab)	48 GB · 25 GB free · ~24 tok/s · $4.64 / 1M tokens
A100 40GB17 GB free	~49 tok/s	$3.47	Deploy on Vast.aifrom $0.61/hr (opens in new tab)	17 GB free · ~49 tok/s · $3.47 / 1M tokens
L40 (48 GB)25 GB free	~27 tok/s	$7.05	Deploy on RunPodfrom $0.69/hr (opens in new tab)	48 GB · 25 GB free · ~27 tok/s · $7.05 / 1M tokens
RTX 5090 (32 GB)9 GB free	~56 tok/s	$3.40	Deploy on RunPodfrom $0.69/hr (opens in new tab)	32 GB · 9 GB free · ~56 tok/s · $3.40 / 1M tokens

Compare GPUs →

Your own private copy of Qwen3 VL 32B Instruct · Q4_K_M · OpenAI-compatible API

Open the llama.cpp listing on RunPod
Open on RunPodServerless (opens in new tab)
New to RunPod? Sign up and add credit first.
2
Fill in the form
Press Deploy and fill in:
Model
Qwen/Qwen3-VL-32B-Instruct-GGUF
Quantization
Q4_K_M
Context Size
16384
After pasting the model press Esc — don’t pick a suggestion. Press Next.
Create the endpoint
Untick 24 GB, keep 48 GB, then Create Endpoint. Right away: Manage → Edit endpoint → Max workers 1 → Save (changing settings later restarts the worker).
The first start downloads the model, usually 5–15 min. After that, replies after a pause come in seconds.
Create an API key
RunPod → Settings → API Keys → Create. Pick “Restricted” with access to this endpoint only.

Use it

Copy the endpoint URL from its page on RunPod (any of its links works) and paste it with the key. Both stay in this browser.

Endpoint URL

API key

Chats and keys stay only in this browser.

https://api.runpod.ai/v2/<ENDPOINT_ID>/openai/v1

Put your endpoint ID in place of <ENDPOINT_ID>. The API key is the one from step 2.

from openai import OpenAI
client = OpenAI(base_url="https://api.runpod.ai/v2/<ENDPOINT_ID>/openai/v1", api_key="rpa_...")
r = client.chat.completions.create(
    model="any",
    messages=[{"role": "user", "content": "Hello!"}])
print(r.choices[0].message.content)

Works with any OpenAI client.

https://api.runpod.ai/v2/<ENDPOINT_ID>/openai/v1

Put your endpoint ID in place of <ENDPOINT_ID>. The API key is the one from step 2.

Free on Windows, macOS, Linux, iOS, Android and the web.

Open the web app ↗ (opens in a new tab)Download ↗ (opens in a new tab)

1. Settings → Model Provider → Add → Add Custom Provider
2. API Mode OpenAI API Compatible; paste the API key and the base URL as API Host
3. Model → Fetch, pick the model
4. New chat → pick the provider → start chatting

Free and open source for Windows, macOS and Linux.

Download Jan ↗ (opens in a new tab)

1.Settings → next to Model Providers press +
2.API format OpenAI-compatible; Base URL is the base URL above (with /v1); API key is your key; Create
3.If no model appears, click + in Models and type any
4.New chat → pick the model → start chatting

How Much VRAM Does Qwen3 VL 32B Instruct Need?

Select a quantization level; the listed hardware adheres to it.

VRAM	+32K	+128K	File	Quality	On RTX 4090 (reference)
9.8 GB	14.8 GB	31 GB	9.3 GB	2-bit · i-quant extreme compression — Importance-weighted 2-bit, extreme compression — significant quality loss	Fits: 64K context · ~67 tok/s
10.6 GB	15.7 GB	31.8 GB	10 GB	2-bit · i-quant extra small — Importance-weighted 2-bit, extra small	Fits: 64K context · ~62 tok/s
11.1 GB	16.1 GB	32.2 GB	10.5 GB	2-bit · i-quant small — Importance-weighted 2-bit, small	Fits: 64K context · ~59 tok/s
11.9 GB	16.9 GB	33 GB	11.6 GB	2-bit · i-quant medium — Importance-weighted 2-bit, medium	Fits: 64K context · ~55 tok/s
13.6 GB	18.6 GB	34.7 GB	13 GB	3-bit · i-quant — Importance-weighted 3-bit	Fits: 64K context · ~48 tok/s
14.0 GB	19 GB	35.1 GB	11.5 GB	2-bit · small K-quant — 2-bit small K-quant	Fits: 32K context · ~47 tok/s
14.4 GB	19.4 GB	35.5 GB	13.7 GB	3-bit · i-quant extra small — Importance-weighted 3-bit, extra small	Fits: 32K context · ~46 tok/s
14.8 GB	19.8 GB	36 GB	14.4 GB	3-bit · i-quant small — Importance-weighted 3-bit, small	Fits: 32K context · ~44 tok/s
14.8 GB	19.8 GB	36 GB	12.3 GB	2-bit · K-quant — 2-bit quantization with K-quant improvements	Fits: 32K context · ~44 tok/s
15.2 GB	20.3 GB	36.4 GB	14.4 GB	3-bit small — 3-bit small quantization	Fits: 32K context · ~43 tok/s
15.7 GB	20.7 GB	36.8 GB	14.9 GB	3-bit · i-quant medium — Importance-weighted 3-bit, medium	Fits: 32K context · ~42 tok/s
16.9 GB	21.9 GB	38 GB	16 GB	3-bit medium — 3-bit medium quantization	Fits: 32K context · ~39 tok/s
17.3 GB	22.3 GB	38.5 GB	18.7 GB	4-bit · legacy — 4-bit legacy quantization	Fits: 32K context · ~38 tok/s
17.7 GB	22.8 GB	38.9 GB	17.3 GB	3-bit large — 3-bit large quantization	Fits: 32K context · ~37 tok/s
18.6 GB	23.6 GB	39.7 GB	17.7 GB	4-bit · i-quant compact — Importance-weighted 4-bit, compact	Fits: 32K context · ~35 tok/s
19.4 GB	24.4 GB	40.5 GB	20.6 GB	4-bit · legacy+ — 4-bit legacy quantization with offset	Fits: 16K context · ~34 tok/s
19.4 GB	24.4 GB	40.5 GB	18.8 GB	4-bit small — 4-bit small quantization	Fits: 16K context · ~34 tok/s
19.4 GB	24.4 GB	40.5 GB	18.7 GB	4-bit · i-quant non-linear — Importance-weighted 4-bit, non-linear	Fits: 16K context · ~34 tok/s
20.6 GB	25.7 GB	41.8 GB	19.8 GB	4-bit medium — 4-bit medium quantization — most popular sweet spot	Fits: 16K context · ~32 tok/s
21.1 GB	26.1 GB	42.2 GB	20.3 GB	4-bit large — 4-bit large quantization	Fits: 16K context · ~31 tok/s
23.6 GB	28.6 GB	44.7 GB	22.6 GB	5-bit small — 5-bit small quantization	Fits: 4K context · ~28 tok/s
24.4 GB	29.4 GB	45.5 GB	23.2 GB	5-bit medium — 5-bit medium quantization — good quality/size tradeoff	Does not fit: needs 26.8 GB
24.8 GB	29.9 GB	46 GB	23.7 GB	5-bit large — 5-bit large quantization	Does not fit: needs 27.2 GB
28.2 GB	33.2 GB	49.3 GB	26.9 GB	6-bit, very good — 6-bit quantization, very good quality	Does not fit: needs 30.5 GB
34.0 GB	39 GB	55.1 GB	34.8 GB	8-bit, near-lossless — 8-bit quantization, near-lossless	Does not fit: needs 36.3 GB
67.3 GB	72.4 GB	88.5 GB	65.5 GB	16-bit · bfloat — Brain floating point 16 — preferred for training	Does not fit: needs 69.7 GB

Selected Q4_K_M

4-bit medium quantization

23.0GBat 16K context

Weights: 20.0 GB
KV cache: 2.7 GB
Overhead: 0.3 GB

Fits RTX 4090 (reference) · 1.0 GB free · ~32 tok/s

≥50 tok/s needs ≥1.59 TB/s · RTX 5090 ~56

Which Hardware Can Run Qwen3 VL 32B Instruct?

Q4_K_M · 20.6 GB

Qwen3 VL 32B Instruct (Q4_K_M) requires 20.6 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 27+ GB is recommended. Using the full 262K context window can add up to 42.6 GB, bringing total usage to 63.3 GB. 7 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 100 devices with unified memory can run Qwen3 VL 32B Instruct, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Studio M1 Max (32 GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M

Typical hardwareQuantVRAMtok/sContext that fits

8–12 GB · RTX 4060, RTX 3060 12GBDoesn't fit — smallest quant needs 12.2 GB8 GB tier →12 GB tier →

16 GB · RTX 5060 Ti 16GBIQ3_XXS · 15.9 GB · ~22 tok/s · 16K15.9~2216K16 GB tier →

24 GB · RTX 4090referenceQ4_K_M · 23 GB · ~32 tok/s · 16K23~3216K24 GB tier →

Mac 32 GB · Mac Mini M4Q4_K_M · 23 GB · ~4 tok/s · 32K23~432KMacs →

Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 23 GB · ~19 tok/s · 128K23~19128KMacs →

Alternatives to Qwen3 VL 32B Instruct

Qwen 3 family · 79 models →

Similar models

6tok/s on RTX 4090

Model	Score	VRAM	tok/s	Why it is similar
Nanbeige4.3 33BNanbeige	—	23.7 GB	~32	same size
Sage A1.5InternScience	—	22 GB	~151	same size
MiniCPM V 4.7 35B A3BOpenBMB	—	22.1 GB	~169	same size
Llm Jp 4.1 32B A3b Thinkingllm-jp	—	20.1 GB	~188	same size
Llm Jp 4.1 33B Thinkingllm-jp	—	22.5 GB	~35	same size
Agnes 3.0 FlashAgnes-AI	—	22.1 GB	~35	same size

Fine-tunes & variants(2)community builds of Qwen3 VL 32B Instruct

Specifications

Publisher: Alibaba
Family: Qwen 3
Parameters: 33.4B
Architecture: Qwen3VLForConditionalGeneration
Context Length: 262,144 tokens
Vocabulary Size: 151,936
Release Date: 2025-10-19
License: Apache 2.0

Resources

Frequently Asked Questions

How much VRAM does Qwen3 VL 32B Instruct need?

Qwen3 VL 32B Instruct requires 20.6 GB of VRAM at Q4_K_M, or 67.3 GB at BF16. Full 262K context adds up to 42.6 GB (63.3 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 33.4B × 4.8 bits ÷ 8 = 20 GB

KV Cache + Overhead ≈ 0.6 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

KV Cache + Overhead ≈ 43.3 GB (at full 262K context)

VRAM usage by quantization

Q4_K_M

20.6 GB

Q4_K_M + full context

63.3 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 4090 run Qwen3 VL 32B Instruct?

Yes, at Q5_K_S (23.6 GB) or lower. Higher quantizations like Q5_K_M (24.4 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.

What's the best quantization for Qwen3 VL 32B Instruct?

For Qwen3 VL 32B Instruct, Q4_K_M (20.6 GB) offers the best balance of quality and VRAM usage. Q4_K_L (21.1 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 9.8 GB.

VRAM requirement by quantization

IQ2_XXS

9.8 GB~53%

IQ3_XS

14.4 GB~73%

Q3_K_L

17.7 GB~86%

Q4_K_M ★

20.6 GB~89%

Q4_K_L

21.1 GB~90%

BF16

67.3 GB~100%

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Qwen3 VL 32B Instruct on a Mac?

Yes — Mac Mini M2 Pro (16 GB) and 133 other Macs can run Qwen3 VL 32B Instruct, the smallest at IQ2_XXS. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 20.6 GB of usable unified memory (RAM minus macOS overhead), from Mac Studio M1 Max (32 GB).

Can I run Qwen3 VL 32B Instruct locally?

Yes — Qwen3 VL 32B Instruct can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Qwen3 VL 32B Instruct?

At Q4_K_M, Qwen3 VL 32B Instruct can reach ~232 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~32 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = (bandwidth GB/s ÷ model GB) × efficiency

Example: NVIDIA B200 → 8000 ÷ 20.6 × 0.65 = ~252 tok/s

Estimated speed at Q4_K_M (20.6 GB)

NVIDIA B200

~252 tok/s

NVIDIA GeForce RTX 4090

~32 tok/s

NVIDIA B300

~252 tok/s

AMD Instinct MI350X

~232 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of Qwen3 VL 32B Instruct?

At Q4_K_M, the download is about 20.01 GB. The full-precision BF16 version is 66.71 GB. The smallest option (IQ2_XXS) is 9.17 GB.

Which GPUs can run Qwen3 VL 32B Instruct?

7 consumer GPUs can run Qwen3 VL 32B Instruct at Q4_K_M. Top options include NVIDIA GeForce RTX 5090, AMD Radeon RX 7900 XTX, NVIDIA GeForce RTX 3090. 1 GPU has plenty of headroom for comfortable inference.

Which devices can run Qwen3 VL 32B Instruct?

100 devices with unified memory can run Qwen3 VL 32B Instruct at Q4_K_M (20.6 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.