AlibabaQwen 3Qwen3VLMoeForConditionalGeneration

Qwen3 VL 235B A22B Instruct — Hardware Requirements & GPU Compatibility

VisionFunctions

Qwen3 VL 235B A22B Instruct is Alibaba's large vision-language model from the Qwen3 series, with 235.7 billion total parameters arranged as a Mixture-of-Experts network that activates about 22.8 billion per token, keeping generation relatively fast even though the full weight set must still fit in memory. It combines text generation with image and video understanding, including document parsing, OCR, and agentic tasks such as operating on-screen GUI elements. At this scale, local inference needs multiple high-VRAM GPUs or a large unified-memory machine, so most people use a hosted endpoint instead. It supports a 262K token context window for long documents, multi-image input, and video. It is released under the Apache 2.0 license, and was published in September 2025, pairing Qwen3's vision-language stack with agentic GUI control and video grounding.

235.7B params · MoE23B active262K contextApache 2.0Sep 2025323.5K downloads431 likes10.6K 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.

Qwen3 VL 235B A22B Instruct runs on a 192 GB+ Mac or a datacenter GPU

Best local pick → Mac Pro M2 Ultra (192 GB) · Q4_K_M · 143.3 GB · ~23 tok/s

Macs from 128 GB (Q4_0 · ~26 tok/s)

Apache 2.0

Not set up for it?

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Qwen3 VL 235B A22B 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 · 96 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 235B A22B Instruct

143.3 GB VRAM at 16K context · runs on one GH200 Grace Hopper Superchip 144 GB

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:235b-a22b-instruct
Official Ollama library build. For the exact Q4_K_M file from unsloth/Qwen3-VL-235B-A22B-Instruct-GGUF: ollama run hf.co/unsloth/Qwen3-VL-235B-A22B-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 paste https://huggingface.co/unsloth/Qwen3-VL-235B-A22B-Instruct-GGUF, then pick Q4_K_M. Or use the CLI (run LM Studio once first):
$lms get https://huggingface.co/unsloth/Qwen3-VL-235B-A22B-Instruct-GGUF@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 unsloth/Qwen3-VL-235B-A22B-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/unsloth/Qwen3-VL-235B-A22B-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 unsloth/Qwen3-VL-235B-A22B-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. Splits the model over 8 GPUs of one server (tensor parallel).
$vllm serve Qwen/Qwen3-VL-235B-A22B-Instruct --max-model-len 16384 --tensor-parallel-size 8
Runs the original weights, not the Q4_K_M GGUF; needs ~518 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 LibraxisAI/Qwen3-VL-235B-A22B-Instruct-mlx-nvfp4
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 LibraxisAI/Qwen3-VL-235B-A22B-Instruct-mlx-nvfp4

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.
$unsloth/Qwen3-VL-235B-A22B-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
B200 (192 GB)CheapestCheapest48 GB free	~128 tok/s	$12.97	Deploy on RunPodfrom $5.98/hr (opens in new tab)	192 GB · 48 GB free · ~128 tok/s · $12.97 / 1M tokens
B300 (288 GB)144 GB free	~128 tok/s	$15.05	Deploy on RunPodfrom $6.94/hr (opens in new tab)	288 GB · 144 GB free · ~128 tok/s · $15.05 / 1M tokens
B200 (192 GB)48 GB free	~128 tok/s	$16.27	Deploy on Vast.aifrom $7.50/hr (opens in new tab)	192 GB · 48 GB free · ~128 tok/s · $16.27 / 1M tokens

Compare GPUs →

Your own private copy of Qwen3 VL 235B A22B Instruct · Q4_K_M · OpenAI-compatible API

The pod’s first start downloads ~141 GB, billed while it loads.

Open the pod on RunPod
Deploy on RunPodH200 · $5.29/hr (opens in new tab)
New to RunPod? Sign up and add credit first. If that GPU is gone, pick another with at least 96 GB.
2
Set the model and your key
Press Set overrides → Environment variables. The first variable is already there; set its value:
Key
LLAMA_ARG_HF_REPO
Value
unsloth/Qwen3-VL-235B-A22B-Instruct-GGUF:Q4_K_M
Press Add Environment variable for your key:
Key
LLAMA_API_KEY
Value
Make up your own key
A long password works. You’ll paste it below as the API key. Without it, anyone who finds the pod’s address can use it.Also set Container disk to 170 GB.Then Set overrides.
Deploy
Press Deploy Pod. Usually ready in 15–60 min.

Use it

RunPod → your pod → Connect → HTTP 8080, copy the link. The key is the one you made up in step 2.

Endpoint URL

API key

Chats and keys stay only in this browser.

https://<POD_ID>-8080.proxy.runpod.net/v1

RunPod → your pod → Connect → HTTP 8080 gives this address. The API key is the one from step 2.

from openai import OpenAI
client = OpenAI(base_url="https://<POD_ID>-8080.proxy.runpod.net/v1", api_key="YOUR_KEY")
r = client.chat.completions.create(
    model="any",
    messages=[{"role": "user", "content": "Hello!"}])
print(r.choices[0].message.content)

Works with any OpenAI client.

https://<POD_ID>-8080.proxy.runpod.net/v1

RunPod → your pod → Connect → HTTP 8080 gives this address. 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 235B A22B Instruct Need?

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

VRAM	+32K	+128K	File	Quality	Runs on
91.8 GB	94.8 GB	104.2 GB	94.6 GB	3-bit · i-quant — Importance-weighted 3-bit	Runs on RTX PRO 6000 Blackwell Max-Q Workstation Edition 96 GB
100.7 GB	103.6 GB	113.1 GB	85.7 GB	2-bit · K-quant — 2-bit quantization with K-quant improvements	Runs on Instinct MI250X 128 GB
103.6 GB	106.6 GB	116 GB	101.4 GB	3-bit small — 3-bit small quantization	Runs on Instinct MI250X 128 GB
115.4 GB	118.4 GB	127.8 GB	112.4 GB	3-bit medium — 3-bit medium quantization	Runs on Instinct MI250X 128 GB
118.3 GB	121.3 GB	130.8 GB	133.1 GB	4-bit · legacy — 4-bit legacy quantization	Runs on Instinct MI250X 128 GB
127.2 GB	130.1 GB	139.6 GB	125.3 GB	4-bit · i-quant compact — Importance-weighted 4-bit, compact	Runs on H200 NVL 141 GB
133.1 GB	136 GB	145.5 GB	147.2 GB	4-bit · legacy+ — 4-bit legacy quantization with offset	Runs on H200 NVL 141 GB
133.1 GB	136 GB	145.5 GB	133.7 GB	4-bit small — 4-bit small quantization	Runs on H200 NVL 141 GB
133.1 GB	136 GB	145.5 GB	132.6 GB	4-bit · i-quant non-linear — Importance-weighted 4-bit, non-linear	Runs on H200 NVL 141 GB
141.9 GB	144.9 GB	154.3 GB	142.2 GB	4-bit medium — 4-bit medium quantization — most popular sweet spot	Runs on GH200 Grace Hopper Superchip 144 GB
162.5 GB	165.5 GB	174.9 GB	161.9 GB	5-bit small — 5-bit small quantization	Runs on B200 192 GB
168.4 GB	171.4 GB	180.8 GB	166.8 GB	5-bit medium — 5-bit medium quantization — good quality/size tradeoff	Runs on B200 192 GB
194.9 GB	197.9 GB	207.3 GB	193 GB	6-bit, very good — 6-bit quantization, very good quality	Runs on Instinct MI325X 256 GB
236.2 GB	239.1 GB	248.6 GB	249.9 GB	8-bit, near-lossless — 8-bit quantization, near-lossless	Runs on Instinct MI325X 256 GB
471.8 GB	474.8 GB	484.3 GB	470.3 GB	16-bit · bfloat — Brain floating point 16 — preferred for training	Runs on 7× H100 SXM 80 GB

Selected Q4_K_M

4-bit medium quantization

143.3GBat 16K context

Weights: 141.4 GB
KV cache: 1.6 GB
Overhead: 0.3 GB

Server-class: ~143.3 GB at 16K context · one GH200 Grace Hopper Superchip 144 GB

Or .

Which Hardware Can Run Qwen3 VL 235B A22B Instruct?

Q4_K_M · 141.9 GB

Qwen3 VL 235B A22B Instruct (Q4_K_M) requires 141.9 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 185+ GB is recommended. Using the full 262K context window can add up to 25.0 GB, bringing total usage to 166.9 GB. No consumer GPU has enough memory. 4 devices with unified memory can run Qwen3 VL 235B A22B Instruct, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Pro M2 Ultra (192 GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M

Consumer GPUs (up to 32 GB) and Macs up to 96 GB: too small for this model (needs ≥ 93.2 GB)8 GB tier →12 GB tier →16 GB tier →24 GB tier →Macs →

How good is Qwen3 VL 235B A22B Instruct?

llmrun Score blends public benchmark boards into one number; 100 = the average open model.

Not enough data yet for an llmrun Score.

A Score needs results on enough independent boards. How the Composite Score is built

Benchmarks

LMArena Text
LMArena · Oct 2026Arena
1413.3
#42/188 open
Arena

Qwen3 VL 235B A22B Instruct benchmark results

Alternatives to Qwen3 VL 235B A22B Instruct

Qwen 3 family · 79 models →

Similar models

Model	Score	VRAM	tok/s	Why it is similar
MiniMax M2.7MiniMax	107.4	139.6 GB	—needs > 24 GB	same size
Command A Plus 05 2026 BF16Cohere	—	132.1 GB	—needs > 24 GB	same size
DeepSeek V2.5 1210DeepSeek	—	142.9 GB	—needs > 24 GB	same size
Laguna M.1poolside	—	138.1 GB	—needs > 24 GB	same size
Llama 3 1 Nemotron Ultra 253B V1NVIDIA	—	564.2 GB	—needs > 24 GB	same size

Specifications

Publisher: Alibaba
Family: Qwen 3
Parameters: 235.7B
Active parameters: 22.8B per token
Architecture: Qwen3VLMoeForConditionalGeneration
Context Length: 262,144 tokens
Vocabulary Size: 151,936
Release Date: 2025-09-22
License: Apache 2.0

Resources

Frequently Asked Questions

How much VRAM does Qwen3 VL 235B A22B Instruct need?

Qwen3 VL 235B A22B Instruct requires 141.9 GB of VRAM at Q4_K_M, or 471.8 GB at BF16. Full 262K context adds up to 25.0 GB (166.9 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 235.7B × 4.8 bits ÷ 8 = 141.4 GB

KV Cache + Overhead ≈ 0.5 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 ≈ 25.5 GB (at full 262K context)

VRAM usage by quantization

Q4_K_M

141.9 GB

Q4_K_M + full context

166.9 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 5090 run Qwen3 VL 235B A22B Instruct?

No — Qwen3 VL 235B A22B Instruct requires at least 91.8 GB at IQ3_XXS, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.

What's the best quantization for Qwen3 VL 235B A22B Instruct?

For Qwen3 VL 235B A22B Instruct, Q4_K_M (141.9 GB) offers the best balance of quality and VRAM usage. Q5_K_S (162.5 GB) provides better quality if you have the VRAM. The smallest option is IQ3_XXS at 91.8 GB.

VRAM requirement by quantization

IQ3_XXS

91.8 GB~70%

Q4_0

118.3 GB~85%

Q4_K_S

133.1 GB~88%

Q4_K_M ★

141.9 GB~89%

Q5_K_M

168.4 GB~92%

BF16

471.8 GB~100%

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Qwen3 VL 235B A22B Instruct on a Mac?

Yes — Mac Pro M2 Ultra (128 GB) and 14 other Macs can run Qwen3 VL 235B A22B Instruct, the smallest at Q4_0. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 141.9 GB of usable unified memory (RAM minus macOS overhead), from Mac Pro M2 Ultra (192 GB).

Can I run Qwen3 VL 235B A22B Instruct locally?

Yes — Qwen3 VL 235B A22B Instruct can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Qwen3 VL 235B A22B Instruct?

At Q4_K_M, Qwen3 VL 235B A22B Instruct can reach ~46 tok/s on AMD Instinct MI350X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = 1000 ÷ (active GB ÷ (bandwidth GB/s × efficiency) × 1000 + layers × routing ms)

Mixture-of-Experts: just the active experts get read for each token, along with a fixed routing cost per layer.

Example: NVIDIA B200 → 13.7 GB active ÷ (8000 × 0.65) = 2.64 ms, plus 94 layers × 0.055 ms = 5.17 ms, so 1000 ÷ 7.81 ms = ~128 tok/s

Estimated speed at Q4_K_M (141.9 GB)

NVIDIA B200

~128 tok/s

NVIDIA B300

~128 tok/s

NVIDIA GH200 Grace Hopper Superchip

~106 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 235B A22B Instruct?

At Q4_K_M, the download is about 141.40 GB. The full-precision BF16 version is 471.34 GB. The smallest option (IQ3_XXS) is 91.32 GB.

Which GPUs can run Qwen3 VL 235B A22B Instruct?

No single consumer GPU has enough VRAM to run Qwen3 VL 235B A22B Instruct at Q4_K_M (141.9 GB). Multi-GPU or professional hardware is required.

Which devices can run Qwen3 VL 235B A22B Instruct?

4 devices with unified memory can run Qwen3 VL 235B A22B Instruct at Q4_K_M (141.9 GB), including Mac Pro M2 Ultra (192 GB), Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Studio M2 Ultra (192 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.