[whatmodel.sh](../../README.md) / Qwen3-VL 235B-A22B Instruct
# Qwen3-VL 235B-A22B Instruct
Alibaba · released 2025-09 · 235B params (22B active) · apache-2.0
**score 63.2** ±0.7 · #169 of 732 · from 40 benchmark results across 5 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Qwen3-VL 235B-A22B Instruct among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃▆█●▅▅▄▄▂▁ #174 of 313* 0.6% 1
coding ▁▁▁▃▅▇█●▅▄▃▁ #180 of 469* 2.8% 2
chat & writing ▁▁▁▃▆▄▆█▆●▇▃ #104 of 402 9.2% 3
math ▁▁▃▅█▆▄●▄▂▁▁ #102 of 431 10% 3
vision & documents ▁▁▂▂▁▃▄▂●▇█▄ #80 of 153 44% 15
agent tasks · #174 of 313 · 0.6% of its score
Beats [o1 (2024-12-17)](o1-2024-12-17.md), [o3-mini](o3-mini.md), [DeepSeek R1 0528](deepseek-r1-0528.md) here, though they rank higher overall. Trails [SmolLM3 3B](smollm3-3b.md), [Jamba 1.5 Mini](jamba-1-5-mini.md), [Jamba 1.5 Large](jamba-1-5-large.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
BFCL v3 (maker) 67.7% ▁▁▁▁▁▃▅█▆●▄▁ beats 137/171 medium · avg 56% 0.6% blunt · at its level · small run
coding · #180 of 469 · 2.8% of its score
Trails [QwQ-32B-Preview](qwq-32b-preview.md), [Ministral 3 8B Reasoning 2512](ministral-3-8b-reasoning.md), [Phi-4-reasoning 14B](phi-4-reasoning.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LiveCodeBench v6 (Feb–May 2025) (… 54.3% ▁▁▃▁●▅▄▇▅▅▃█ beats 7/38 hard · avg 30% 1.5% at its level · small run
MultiPL-E (maker) 86.1% ▃··▄·▅▅▇▄█▃● beats 29/31 easy · avg 67% 1.3% small run
chat & writing · #104 of 402 · 9.2% of its score
Beats [LongCat 2.0 (1.6T-A48B)](longcat-2-0.md), [Inkling Small (266B-A12B)](inkling-small.md), [GPT-5.4 nano](gpt-5-4-nano.md) here, though they rank higher overall. Trails [Mistral Large 3 (675B-A41B)](mistral-large-3.md), [Gemma 4 12B](gemma-4-12b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LMArena (text) 1413.3 ▁▁▁▂▃▃▃▆▆●█▃ beats 261/385 avg 1260 5.2% blunt
Arena-Hard v2 aggregate (maker) 77.4% ▁▁··▄▃▁▃█▁●▇ beats 17/22 medium · avg 37% 3.0% sharp · at its level · small run
IFEval (aggregation unspecified) … 87.8% ▁▂▄▂▂▄▆●█▅▂▁ beats 92/145 easy · avg 77% 1.0% small run
math · #102 of 431 · 10% of its score
Beats [Claude Sonnet 4.6](claude-sonnet-4-6.md), [MiniMax-M3](minimax-m3.md), [Nemotron 3 Ultra 550B-A55B](nemotron-3-ultra-550b-a55b.md) here, though they rank higher overall. Trails [Gemini 2.5 Flash](gemini-2-5-flash.md), [GLM-4.5-Air](glm-4-5-air.md), [Granite 4.2 30B](granite-4-2-30b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
MathVista (testmini) (maker) 84.9% ▃··▃▅·▃█▅▅▅● beats 15/16 medium · avg 69% 3.7% typical
HMMT 2025 (edition unspecified) (… 57.4% ▃█●▅▇▅▁·▁··▁ beats 14/32 hard · avg 19% 3.3% sharp · at its level · small run
AIME 2025 (maker) 74.7% ▁·▁▁▁▂▅▆●▇▅▁ beats 75/133 hard · avg 32% 3.2% sharp · at its level · small run
vision & documents · #80 of 153 · 44% of its score
Beats [DeepSeek V4 Flash Vision (Exp)](deepseek-v4-flash-vision-exp.md), [GPT-5.4 nano](gpt-5-4-nano.md), [GPT-5 mini](gpt-5-mini.md) here, though they rank higher overall. Trails [Gemma 4 12B](gemma-4-12b.md), [Qwen3.5 9B](qwen3-5-9b.md), [GPT-4.1](gpt-4-1.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
CharXiv (reasoning questions) (ma… 62.1% ▄▂▄▆▄●▆··▄█▂ beats 11/21 medium · avg 38% 8.4% at its level
MMMU (val) (maker) 78.7% ▂···▆▂▂▂▄▄▄● beats 16/16 medium · avg 61% 6.1% at its level
MMStar (maker) 78.4% ▂··▄▂▂·▄█▂▄● beats 15/16 medium · avg 65% 4.2% at its level
MMMU Pro (maker) 68.1% ▃·█·▅▃▃▃▅▅▃● beats 16/16 medium · avg 46% 4.0% at its level
CharXiv descriptive questions (ma… 89.4% ▄··▂▂▂·▂▄▄▂● beats 11/14 easy · avg 74% 4.0% typical
MathVista (testmini) (maker) 84.9% ▃··▃▅·▃█▅▅▅● beats 15/16 medium · avg 69% 3.7% typical
Arena vision (style controlled) 1215.1 ▁▁▂▂▂▂▃▄▅●█▇ beats 80/152 avg 1061 3.5% blunt
ChartArena chart parsing (English… 46.0% ▁·▁·▃▃▁▇▃●·▄ beats 17/22 hard · avg 27% 2.8% blunt · at its level
HallusionBench (maker) 63.2% ▃·▃·▅▃▅▅▃●█▅ beats 11/16 medium · avg 54% 1.9% blunt · at its level
RealWorldQA (maker) 79.2% ▅▃·▃▅▅·█·▃▅● beats 16/16 easy · avg 70% 1.2% blunt
ChartQA (maker) 90.3% ▂··▂·▂▃·▅▂▆● beats 16/17 easy · avg 85% 1.2% blunt
ScreenSpot-Pro (maker) 62.0% ▂····▂▄··▂█● beats 11/11 medium · avg 46% 1.1% blunt · at its level
DocVQA (maker) 97.1% ▃··▅▃▅▃█▃█▅● beats 17/17 very easy · avg 94% 1.0% blunt · maxed out
CC-OCR (maker) 82.2% ▃··▃▅·▅▅·█▅● beats 13/13 easy · avg 74% 0.8% blunt
OCRBench (maker) 92.0% ▅█·▃▃█▃▅▅▃·● beats 16/16 easy · avg 84% 0.4% blunt
other tests (17)
test result vs the others in its run difficulty counts why
GPQA (subset unspecified) (maker) 74.3% ▂····▂·▂▂█▆● beats 12/14 hard · avg 49% 7.1% sharp · at its level
WebCompass (overall) 41.1% ▅·▃▃●···▃·▃█ beats 4/9 hard · avg 18% 5.2% at its level
LVBench (maker) 67.7% ▅··▃·▅▃▃·█·● beats 10/10 hard · avg 51% 2.9% at its level
AndroidWorld (maker) 63.7% ▃··▃▅▃▃▃▃▃·● beats 11/11 medium · avg 43% 2.8% at its level
Video-MME (without subtitles) (ma… 79.2% ▃··█··▅█▃▃▅● beats 13/13 medium · avg 66% 2.4% blunt · at its level
MMLU-Pro (maker) 81.8% ▁▁·▁▁▁▃▂▄▅●▁ beats 137/188 medium · avg 57% 1.9% small run
SimpleQA (maker) 51.9% ▄▄▄▆█▆███·▂● beats 29/32 hard · avg 18% 1.8% at its level · small run
INCLUDE (maker) 80.0% ▂····▂▂▄▄▄●▄ beats 29/36 medium · avg 58% 1.7% at its level · small run
MMLU-Redux (maker) 92.2% ▁▁▃▁▂▃▄▅▇●█▄ beats 45/63 easy · avg 80% 1.7% small run
ChineseSimpleQA (maker) 83.4% ▄··▄▄▄··▄··● beats 5/6 medium · avg 59% 1.3% small run
Creative Writing v3 (maker) 86.5% ▂····▂▆··▆█● beats 12/14 easy · avg 69% 1.3% small run
SuperGPQA (maker) 60.4% ▂▁▂▂▅▆▃▇▇●█▃ beats 34/46 medium · avg 38% 1.1% at its level · small run
MMLU-ProX (maker) 77.8% ▁·▁▁▁▂▁▃▃●▅▄ beats 17/36 medium · avg 58% 1.1% small run
MVBench (maker) 76.5% ▂▂▂·▂█··▆▄·● beats 13/13 medium · avg 67% 1.1% blunt · at its level
WritingBench (maker) 85.5% ▁·▁▂▁··▅█●▂▁ beats 16/20 easy · avg 69% 1.1% small run
LiveBench 2024-11-25 (maker) 74.8% ▄··▂·▂·▂●▆▂▂ beats 5/13 medium · avg 57% 1.0% small run
Multi-IF (aggregation unspecified… 76.3% ▁▁▁▁▁▁▃▄●▄▁▂ beats 21/41 medium · avg 61% 0.9% small run
source results share of the score
Qwen (submitted / maker-reported) 18 54% ▇▇▇▇▇▇▇
Qwen3-VL-235B-A22B-Instruct model card … 18 29% ▇▇▇
LMArena leaderboard dataset 2 9% ▇
WebCompass: Towards Multimodal Web Codi… 1 5% ▇
ChartArena: Benchmarking Chart Parsing … 1 3% ▇
prior: pull toward the middle 0% ▇
weighted average = 63.2
Licences: Qwen3-VL-235B-A22B-Instruct model card ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · WebCompass: Towards Multimodal Web Coding Evaluation for Code Language Models ([CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/)) · ChartArena: Benchmarking Chart Parsing across Languages ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/))
## How to run it
### llama.cpp (recommended)
1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md)
```sh
# macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
# Windows
irm https://llama.app/install.ps1 | iex
```
2. Run Qwen3-VL 235B-A22B Instruct (Q4_K_M, needs ~142.2 GB, from [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen3-VL-235B-A22B-Instruct-GGUF/tree/cbf7709dad5de4632836beb8969d51f1eff0f48b))
```sh
llama cli -hf Qwen/Qwen3-VL-235B-A22B-Instruct-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf Qwen/Qwen3-VL-235B-A22B-Instruct-GGUF:Q4_K_M
```
Other versions: Q8_0 (~249.9 GB) `llama cli -hf Qwen/Qwen3-VL-235B-A22B-Instruct-GGUF:Q8_0`
### LM Studio
1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download)
2. Run Qwen3-VL 235B-A22B Instruct (Q4_K_M, needs ~142.2 GB, from [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen3-VL-235B-A22B-Instruct-GGUF/tree/cbf7709dad5de4632836beb8969d51f1eff0f48b))
```sh
lms get https://huggingface.co/Qwen/Qwen3-VL-235B-A22B-Instruct-GGUF@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: Q8_0 (~249.9 GB) `lms get https://huggingface.co/Qwen/Qwen3-VL-235B-A22B-Instruct-GGUF@Q8_0 -y` · mxfp4 (~126 GB, Mac only) `lms get https://huggingface.co/LibraxisAI/Qwen3-VL-235B-A22B-Instruct-mlx-mxfp4 -y`
### Ollama
1. Install Ollama. [Download it (macOS / Windows)](https://ollama.com/download)
```sh
# Linux
curl -fsSL https://ollama.com/install.sh | sh
```
2. Run Qwen3-VL 235B-A22B Instruct (Q4_K_M, needs ~141.5 GB, from [ollama.com](https://ollama.com/library/qwen3-vl/tags))
```sh
ollama run qwen3-vl:235b-a22b-instruct-q4_K_M
```
Other versions: Q8_0 (~282.5 GB) `ollama run qwen3-vl:235b-a22b-instruct-q8_0`
### MLX (Mac)
1. Install MLX (Mac). Apple silicon Macs only, text only.
```sh
# with uv
uv tool install mlx-lm
# or
pip install mlx-lm
```
2. Run Qwen3-VL 235B-A22B Instruct (mxfp4, needs ~126 GB, Mac only, from [huggingface.co/LibraxisAI](https://huggingface.co/LibraxisAI/Qwen3-VL-235B-A22B-Instruct-mlx-mxfp4/tree/b40acc17e844f554012292199bb3ad4ae54a5b1b))
```sh
mlx_lm.chat --model LibraxisAI/Qwen3-VL-235B-A22B-Instruct-mlx-mxfp4
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model LibraxisAI/Qwen3-VL-235B-A22B-Instruct-mlx-mxfp4
```
Model repository and documentation: [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen3-VL-235B-A22B-Instruct)
[interactive version](https://whatmodel.sh/model/qwen3-vl-235b-a22b-instruct) · [JSON](https://whatmodel.sh/api?format=json&all=1)