whatmodel.sh / Qwen3-VL 235B-A22B Instruct

Qwen3-VL 235B-A22B Instruct

+ contribute

Qwen3-VL 235B-A22B Instruct is an open-weights language model from Alibaba, released September 2025. Its published size is 235B parameters (22B active per token) and a 256k-token context window. Licence: apache-2.0.

Catalogue identity notes

Arena entry 'qwen3-vl-235b-a22b-instruct' (text, vision). Maker repo Qwen/Qwen3-VL-235B-A22B-Instruct @710c138: licence tag apache-2.0. Parameters: 235B total / 22B active per the model name (Qwen3-VL-235B-A22B); HF safetensors metadata 235.7B. Released 2025-09: month the maker's Hugging Face repo was created (createdAt 2025-09-22).

63.7±0.8#160 of 719 LLMsfrom 39 benchmark results across 4 sources

$ how it compares

The LLMs ranked either side of Qwen3-VL 235B-A22B Instruct, out of 828 with a score, then more from Alibaba.

#	model	score	where	RAM	released
181	Gemma 4 26B-A4B Google	64±1	api $0.14/M +local	~18GB	2026-04
182	Kimi K2 Instruct (0905) Moonshot AI	64±2	local	~298GB	2025-09
183	Qwen3-VL 235B-A22B Instruct Alibaba	64±1	local	~134GB	2025-09
184	Nova 2 Pro (Preview) Amazon	64±2	api $3.78/M		2025-11
185	Grok 3 Mini Beta SpaceXAI (xAI)	64±1	comparison only		2025-04
	more from Alibaba
18	Qwen3.8 Max (2.4T-A95B) Alibaba	78±1	api $3.00/M +local	~380GB	2026-08
34	Qwen3.8 Max (0902) Alibaba	76±1	api $3.00/M		2026-09
37	Qwen3.8-Flash-Next Alibaba	75±1	local	~59GB	2026-08

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #167 of 300 * · 0.6% of its score · 1 test

Beats o1 (2024-12-17), o3-mini, DeepSeek R1 0528 here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Mini, Jamba 1.5 Large here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
BFCL v3 lab	67.7% ✎	▁▁▁▁▁▃▅█▆▅▄▁ beats 136/170	medium · avg 57%	▇ 0.6%	blunt · at its level · small run

coding▁▁▁▃▆██▇▆▄▃▁ #171 of 458 * · 3.2% of its score · 2 tests

Beats Qwen3-Coder-Next (80B-A3B) here, though they rank higher overall. Trails QwQ-32B-Preview, Ministral 3 8B Reasoning 2512, Phi-4-reasoning 14B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
LiveCodeBench v6 (Feb–May 2025) lab	54.3% ✎	▁▁▃▁█▅▄▇▅▅▃█ beats 7/38	hard · avg 31%	▇ 1.8%	at its level · small run
MultiPL-E lab	86.1% ✎	▃··▄·▅▅▇▄█▃▄ beats 29/31	easy · avg 67%	▇ 1.4%	small run

chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #97 of 393 · 10% of its score · 3 tests

Beats LongCat 2.0 (1.6T-A48B), Inkling Small (266B-A12B), GPT-5.4 nano here, though they rank higher overall. Trails Mistral Large 3 (675B-A41B), Gemma 4 12B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
LMArena (text)	1413.4 ✎	▁▁▁▂▃▃▃▆▆▇█▃ beats 260/379	avg 1268	▇▇ 5.1%	blunt
Arena-Hard v2 aggregate lab	77.4% ✎	▁▁··▄▃▁▃█▁▃▇ beats 17/22	medium · avg 36%	▇ 3.4%	sharp · at its level · small run
IFEval (aggregation unspecified) lab	87.8% ✎	▁▁▃▁▂▄▄▆█▃▂▁ beats 92/141	easy · avg 77%	▇ 1.1%	small run

math▁▂▃▅▆▅▄█▄▁▁▁ #112 of 398 · 11% of its score · 3 tests

Beats Claude Sonnet 4.6, MiniMax-M3, MiMo-V2.5 (310B-A15B) here, though they rank higher overall. Trails Gemini 2.5 Flash, Qwen3-VL 32B Thinking, gpt-oss-120b here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
MathVista (testmini) lab	84.9% ✎	▃··▃▅·▃█▅▅▅█ beats 15/16	medium · avg 69%	▇▇ 4.0%	typical
HMMT 2025 (edition unspecified) lab	57.4% ✎	▃█▇▅▇▅▁·▁··▁ beats 14/32	hard · avg 19%	▇▇ 3.8%	sharp · at its level · small run
AIME 2025 lab	74.7% ✎	▁·▁▁▁▂▅▆█▇▅▁ beats 75/133	hard · avg 34%	▇ 3.4%	sharp · at its level · small run

vision & documents▁▁▃▂▁▃▄▂▆▇█▄ #74 of 147 · 45% of its score · 15 tests

Beats DeepSeek V4 Flash Vision (Exp), GPT-5.4 nano, GPT-5 mini here, though they rank higher overall. Trails GPT-4.1, Gemma 4 12B, Qwen3.5 9B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
CharXiv (reasoning questions) lab	62.1% ✎	▄▂▄▆▄▄▆··▄█▂ beats 11/21	medium · avg 39%	▇▇▇ 8.4%	at its level
MMMU (val) lab	78.7% ✎	▂···▆▂▂▂▄▄▄█ beats 16/16	medium · avg 61%	▇▇▇ 6.4%	at its level
MMStar lab	78.4% ✎	▂··▄▂▂·▄█▂▄▆ beats 15/16	medium · avg 65%	▇▇ 4.3%	at its level
MMMU Pro lab	68.1% ✎	▃·█·▅▃▃▃▅▅▃█ beats 16/16	medium · avg 46%	▇▇ 4.2%	at its level
CharXiv descriptive questions lab	89.4% ✎	▄··▂▂▂·▂▄▄▂█ beats 11/14	easy · avg 74%	▇▇ 4.0%	typical
MathVista (testmini) lab	84.9% ✎	▃··▃▅·▃█▅▅▅█ beats 15/16	medium · avg 69%	▇▇ 4.0%	typical
Arena vision (style controlled)	1215.1 ✎	▁▁▂▂▂▂▄▅▅▆██ beats 79/145	avg 1068	▇ 3.5%	blunt
ChartArena chart parsing (English, mAP_high average)	46.0% ✎	▁·▁·▃▃▁▇▃█·▄ beats 17/22	hard · avg 27%	▇ 2.8%	blunt · at its level

7 more tests

test	result	vs the others in its run	difficulty	counts	why
HallusionBench lab	63.2% ✎	▃·▃·▅▃▅▅▃▅█▅ beats 11/16	medium · avg 55%	▇ 1.9%	blunt · at its level
RealWorldQA lab	79.2% ✎	▅▃·▃▅▅·█·▃▅█ beats 16/16	easy · avg 70%	▇ 1.3%	blunt
ChartQA lab	90.3% ✎	▂··▂·▂▃·▅▂▆█ beats 16/17	easy · avg 85%	▇ 1.2%	blunt
ScreenSpot-Pro lab	62.0% ✎	▂····▂▄··▂█▆ beats 11/11	medium · avg 46%	▇ 1.2%	blunt · at its level
DocVQA lab	97.1% ✎	▃··▅▃▅▃█▃█▅▅ beats 17/17	very easy · avg 94%	▇ 1.0%	blunt · maxed out
CC-OCR lab	82.2% ✎	▃··▃▅·▅▅·█▅▃ beats 13/13	easy · avg 74%	▇ 0.8%	blunt
OCRBench lab	92.0% ✎	▅█·▃▃█▃▅▅▃·▃ beats 16/16	easy · avg 84%	▇ 0.4%	blunt

other tests16 not in a task · 34% of its score

test	result	vs the others in its run	difficulty	counts	why
GPQA (subset unspecified) lab	74.3% ✎	▂····▂·▂▂█▆█ beats 12/14	hard · avg 49%	▇▇▇ 7.4%	sharp · at its level
LVBench lab	67.7% ✎	▅··▃·▅▃▃·█·▃ beats 10/10	hard · avg 51%	▇ 3.0%	at its level
AndroidWorld lab	63.7% ✎	▃··▃▅▃▃▃▃▃·█ beats 11/11	medium · avg 43%	▇ 3.0%	blunt · at its level
Video-MME (without subtitles) lab	79.2% ✎	▃··█··▅█▃▃▅▃ beats 13/13	medium · avg 66%	▇ 2.5%	blunt · at its level
MMLU-Pro lab	81.8% ✎	▁▁·▁▁▁▃▂▄▅█▁ beats 137/188	medium · avg 58%	▇ 2.1%	small run
SimpleQA lab	51.9% ✎	▄▄▄▆█▆███·▂█ beats 29/32	hard · avg 17%	▇ 2.0%	at its level · small run
INCLUDE lab	80.0% ✎	▂····▂▂▄▄▄█▄ beats 29/36	medium · avg 58%	▇ 2.0%	at its level · small run
MMLU-Redux lab	92.2% ✎	▁▁▃▁▂▃▄▅▇▅█▄ beats 45/63	easy · avg 80%	▇ 1.9%	small run

8 more tests

test	result	vs the others in its run	difficulty	counts	why
ChineseSimpleQA lab	83.4% ✎	▄··▄▄▄··▄··█ beats 5/6	medium · avg 58%	▇ 1.6%	small run
Creative Writing v3 lab	86.5% ✎	▂····▂▆··▆█▆ beats 12/14	easy · avg 69%	▇ 1.5%	small run
SuperGPQA lab	60.4% ✎	▂▁▂▂▅▆▃▇▇▇█▃ beats 34/46	medium · avg 38%	▇ 1.3%	at its level · small run
MMLU-ProX lab	77.8% ✎	▁·▁▁▁▂▁▃▂█▅▄ beats 16/35	medium · avg 58%	▇ 1.3%	small run
WritingBench lab	85.5% ✎	▁·▁▂▁··▅█▃▂▁ beats 16/20	easy · avg 69%	▇ 1.2%	small run
LiveBench 2024-11-25 lab	74.8% ✎	▄··▂·▂·▂█▆▂▂ beats 5/13	medium · avg 57%	▇ 1.1%	small run
MVBench lab	76.5% ✎	▂▂▂·▂█··▆▄·▃ beats 13/13	medium · avg 67%	▇ 1.1%	blunt · at its level
Multi-IF (aggregation unspecified) lab	76.3% ✎	▁▁▁▁▁▁▃▄█▄▁▂ beats 21/41	medium · avg 61%	▇ 1.0%	blunt · small run

$ where to run it

llama.cpp · recommended

version

Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
Windows
irm https://llama.app/install.ps1 | iex
other ways to install ↗
Run Qwen3-VL 235B-A22B Instruct · needs ~142.2 GB · from huggingface.co/Qwen (official)
chat in the terminal
llama cli -hf Qwen/Qwen3-VL-235B-A22B-Instruct-GGUF:Q4_K_M
Or start a local server and open the address it prints (http://localhost:9931)
llama serve -hf Qwen/Qwen3-VL-235B-A22B-Instruct-GGUF:Q4_K_M

LM Studio

version

Install LM Studio · Open the app once before using the lms command.
download it (macOS / Windows / Linux) ↗
Run Qwen3-VL 235B-A22B Instruct · needs ~142.2 GB · from huggingface.co/Qwen (official)
open in LM Studio ↗
Or download it in the terminal
lms get https://huggingface.co/Qwen/Qwen3-VL-235B-A22B-Instruct-GGUF@Q4_K_M -y
Then run lms ls to see its name, and lms chat followed by that name

Ollama

version

Install Ollama
download it (macOS / Windows) ↗
Linux
curl -fsSL https://ollama.com/install.sh | sh
Run Qwen3-VL 235B-A22B Instruct · needs ~141.5 GB · from ollama.com
chat in the terminal
ollama run qwen3-vl:235b-a22b-instruct-q4_K_M

MLX (Mac)

Install MLX (Mac) · Apple silicon Macs only, text only.
with uv
uv tool install mlx-lm
or
pip install mlx-lm
Run Qwen3-VL 235B-A22B Instruct · needs ~126 GB, Mac only · from huggingface.co/LibraxisAI
chat in the terminal
mlx_lm.chat --model LibraxisAI/Qwen3-VL-235B-A22B-Instruct-mlx-mxfp4
Or start an OpenAI-compatible server at http://localhost:8080/v1
mlx_lm.server --model LibraxisAI/Qwen3-VL-235B-A22B-Instruct-mlx-mxfp4

Official weights and documentation: huggingface.co/Qwen ↗

$ how the score was worked out · Qwen (Alibaba) vendor-reporte… 56%, Qwen3-VL-235B-A22B-Instruct m… 32%, LMArena leaderboard dataset 9% and 1 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

source	results	share of the score	licence
Qwen (Alibaba) vendor-reported benchmark scores for its own models (qwen.ai launch posts, Qwen2.5-VL and Qwen3-VL technical reports, Qwen3.8-Flash-Next card) · lab-reported	18	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 56%	Factual vendor-reported score (not licensed content)
Qwen3-VL-235B-A22B-Instruct model card · lab-reported	18	▇▇▇▇▇▇▇▇▇▇ 32%	Apache-2.0
LMArena leaderboard dataset	2	▇▇▇ 9%	CC BY 4.0
ChartArena: Benchmarking Chart Parsing across Languages, Scenarios, and Formats (arXiv:2606.01348v3)	1	▇ 3%	CC BY 4.0
prior · pull toward the middle		▇ 0%
weighted average		= 63.7

$ improve this page

Update model information Add a place to run it Add a test result Propose a benchmark Correct a benchmark score Edit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated 2026-10-11 02:20 UTC · sources and licences · json