[whatmodel.sh](../../README.md) / Qwen3-VL 2B Instruct
# Qwen3-VL 2B Instruct
Alibaba · released 2025-10 · 2.13B params · apache-2.0
**score 44.5** ±0.7 · #487 of 732 · from 41 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 2B Instruct among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃●█▄▅▅▄▄▂▁ #248 of 313* 1.9% 1
coding ▁▁▁●▅▇█▇▅▄▃▁ #424 of 469* 2.5% 1
chat & writing ▁▁▁▃▆▄●█▆▇▇▃ #287 of 402* 4.3% 1
math ▁▁▃▅●▆▄█▄▂▁▁ #304 of 431* 8.8% 2
vision & documents ▁▁▂●▁▃▄▂▆▇█▄ #137 of 153 53% 19
agent tasks · #248 of 313 · 1.9% of its score
Beats [o3-mini](o3-mini.md), [Llama 4 Maverick (402B-A17B)](llama-4-maverick.md), [DeepSeek R1 0528 Qwen3 8B](deepseek-r1-0528-qwen3-8b.md) here, though they rank higher overall. Trails [Granite 4.0 1B](granite-4-0-1b.md), [SmolLM3 3B](smollm3-3b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
BFCL v3 (maker) 55.4% ●▄··▄█·█▄█▄█ beats 0/13 medium · avg 59% 1.9% blunt · at its level
coding · #424 of 469 · 2.5% of its score
Beats [DeepSeek R1 Distill Qwen 7B](deepseek-r1-distill-qwen-7b.md), [Qwen3.5 2B](qwen3-5-2b.md), [DeepSeek R1 Distill Llama 8B](deepseek-r1-distill-llama-8b.md) here, though they rank higher overall. Trails [Llama 3.2 1B Instruct](llama-3-2-1b.md), [SmolLM2 1.7B Instruct](smollm2-1-7b.md), [Granite 3B Code Base (2K)](granite-3b-code-base-2k.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LiveCodeBench v6 (window unspecif… 20.3% ●▅█▅▅▃▅···▃█ beats 0/16 hard · avg 35% 2.5% typical
chat & writing · #287 of 402 · 4.3% of its score
Beats [Ministral 3 14B Reasoning 2512](ministral-3-14b-reasoning.md), [Ministral 3 8B](ministral-3-8b.md), [Gemma 4 E4B](gemma-4-e4b.md) here, though they rank higher overall. Trails [EXAONE 3.5 2.4B Instruct](exaone-3-5-2-4b.md), [LFM2 700M](lfm2-700m.md), [Flan-T5 XXL](flan-t5-xxl.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
IFEval (aggregation unspecified) … 68.2% ●··▃···██▅▃▃ beats 0/11 easy · avg 77% 4.3% at its level · several runs
math · #304 of 431 · 8.8% of its score
Beats [Molmo 2 8B](molmo-2-8b.md), [Llama 3.1 Nemotron 70B Instruct](llama-3-1-nemotron-70b.md), [GPT-4 (0613)](gpt-4-0613.md) here, though they rank higher overall. Trails [DeepSeek R1 Distill Qwen 1.5B](deepseek-r1-distill-qwen-1-5b.md), [Gemma 7B (base)](gemma-7b-base.md), [Mistral 7B v0.1 (base)](mistral-7b-v0-1.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
MathVista (testmini) (maker) 61.3% ▃··▃●·▃█▅▅▅█ beats 2/16 medium · avg 69% 5.3% at its level
AIME 2025 (maker) 22.2% ●·▃▃█··▃▅▃·█ beats 0/12 medium · avg 44% 3.4% sharp · at its level
vision & documents · #137 of 153 · 53% of its score
Beats [Gemma 4 E2B](gemma-4-e2b.md), [Gemma 3 4B](gemma-3-4b.md), [LLaVA-OneVision Qwen2 72B (ov-sft)](llava-onevision-qwen2-72b-ov.md) here, though they rank higher overall.
test result vs the others in its run difficulty counts why
CharXiv descriptive questions (ma… 62.3% ●··▂▂▂·▂▄▄▂█ beats 1/14 easy · avg 74% 11% at its level · several runs
CharXiv (reasoning questions) (ma… 26.8% ●▂▄▆▄▄▆··▄█▂ beats 0/21 medium · avg 38% 6.6% at its level
MMMU (val) (maker) 53.4% ▂···●▂▂▂▄▄▄█ beats 2/16 medium · avg 61% 6.4% at its level
MMStar (maker) 58.3% ▂··●▂▂·▄█▂▄▆ beats 2/16 medium · avg 65% 6.1% at its level · several runs
MathVista (testmini) (maker) 61.3% ▃··▃●·▃█▅▅▅█ beats 2/16 medium · avg 69% 5.3% at its level
MMMU Pro (maker) 36.5% ▃·●·▅▃▃▃▅▅▃█ beats 1/16 medium · avg 46% 3.0% at its level
HallusionBench (maker) 51.4% ▃·▃·●▃▅▅▃▅█▅ beats 3/16 medium · avg 54% 2.6% blunt · at its level · several runs
ChartQA (maker) 79.1% ▂··●·▂▃·▅▂▆█ beats 1/17 easy · avg 85% 1.9% blunt · several runs
RealWorldQA (maker) 63.9% ●▃·▃▅▅·█·▃▅█ beats 1/16 easy · avg 70% 1.9% blunt · at its level · several runs
DocVQA (maker) 93.3% ▃··●▃▅▃█▃█▅▅ beats 2/17 very easy · avg 94% 1.8% blunt · maxed out
ChartArena chart parsing (English… 27.0% ▁·▁·▃●▁▇▃█·▄ beats 5/22 hard · avg 27% 1.6% blunt
ScreenSpot-Pro (maker) 48.5% ▂····▂●··▂█▆ beats 2/11 medium · avg 46% 1.0% blunt · at its level
CC-OCR (maker) 72.8% ▃··●▅·▅▅·█▅▃ beats 1/13 easy · avg 74% 1.0% blunt · at its level
MMBench-EN dev v1.1 74.4% █·····▃··●·▃ beats 4/5 easy · avg 79% 0.9% blunt · at its level
BEAR-Bench (bilingual RU/EN enter… 14.3% ▃●·▅·▃·█·█▅▅ beats 1/14 hard · avg 24% 0.9% typical
OCRBench (maker) 85.8% ▅█·▃▃●▃▅▅▃·▃ beats 7/16 easy · avg 84% 0.8% blunt · several runs
CountBench 84.8% ▄··▄··█··▄·● beats 5/5 easy · avg 84% 0.4% noisy
AI2D test 71.3% ●···▄▄···▄·▄ beats 1/5 easy · avg 81% 0.2% blunt · at its level
PureDocBench document parsing (Av… 62.1% ▂·▁·●▂▅█·▁▂▂ beats 3/22 medium · avg 62% 0.1% blunt · at its level
other tests (10)
test result vs the others in its run difficulty counts why
MMLU-Pro (maker) 49.0% ●·▄·▄▄▄▄█▄██ beats 0/12 medium · avg 63% 9.2% at its level · several runs
MMLU-Redux (maker) 66.5% ●··▃·▃·█▃▅▃▅ beats 0/11 easy · avg 79% 6.7% at its level
GPQA (subset unspecified) (maker) 42.0% ▂····●·▂▂█▆█ beats 1/14 hard · avg 49% 4.6% sharp
SuperGPQA (maker) 24.3% ●··▄·▄▄▄█▄██ beats 0/11 medium · avg 37% 2.9% at its level
Video-MME (without subtitles) (ma… 61.9% ▃··●··▅█▃▃▅▃ beats 1/13 medium · avg 66% 2.5% blunt · at its level
AndroidWorld (maker) 36.4% ●··▃▅▃▃▃▃▃·█ beats 0/11 medium · avg 43% 2.4% at its level
MMLU 63.0% ▄·····▄▄··●█ beats 3/5 medium · avg 69% 2.4% at its level
LVBench (maker) 47.4% ●··▃·▅▃▃·█·▃ beats 0/10 hard · avg 51% 2.0% at its level
MVBench (maker) 61.7% ●▂▂·▂█··▆▄·▂ beats 0/13 medium · avg 67% 1.0% blunt · at its level
Multi-IF (aggregation unspecified) 52.3% ▄·▄··█●····▄ beats 4/5 medium · avg 57% 0.8% at its level
source results share of the score
Qwen (submitted / maker-reported) 25 84% ▇▇▇▇▇▇▇▇▇▇
CohereLabs/North-Micro-Vision-Instruct … 13 13% ▇▇
ChartArena: Benchmarking Chart Parsing … 1 2% ▇
BEAR-Bench: A Bilingual Enterprise and … 1 1% ▇
How Far Is Document Parsing from Solved… 1 0% ▇
prior: pull toward the middle 0% ▇
weighted average = 44.5
Licences: CohereLabs/North-Micro-Vision-Instruct model card ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0)) · ChartArena: Benchmarking Chart Parsing across Languages ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · BEAR-Bench: A Bilingual Enterprise and Academic Reasoning Benchmark for Multimodal Models ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · How Far Is Document Parsing from Solved? PureDocBench: A Source-Traceable Benchmark across Clean ([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 2B Instruct (Q4_K_M, needs ~1.1 GB, from [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct-GGUF/tree/52d6c8ffea26cc873ac5ad116f8631268d7eb503))
```sh
llama cli -hf Qwen/Qwen3-VL-2B-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-2B-Instruct-GGUF:Q4_K_M
```
Other versions: Q8_0 (~1.8 GB) `llama cli -hf Qwen/Qwen3-VL-2B-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 2B Instruct (Q4_K_M, needs ~1.1 GB, from [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct-GGUF/tree/52d6c8ffea26cc873ac5ad116f8631268d7eb503))
```sh
lms get https://huggingface.co/Qwen/Qwen3-VL-2B-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 (~1.8 GB) `lms get https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct-GGUF@Q8_0 -y` · 4bit (~1.8 GB, Mac only) `lms get https://huggingface.co/mlx-community/Qwen3-VL-2B-Instruct-4bit -y` · 8bit (~2.6 GB, Mac only) `lms get https://huggingface.co/lmstudio-community/Qwen3-VL-2B-Instruct-MLX-8bit -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 2B Instruct (Q4_K_M, needs ~1.9 GB, from [ollama.com](https://ollama.com/library/qwen3-vl/tags))
```sh
ollama run qwen3-vl:2b-instruct
```
Other versions: Q8_0 (~2.6 GB) `ollama run qwen3-vl:2b-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 2B Instruct (4bit, needs ~1.8 GB, Mac only, from [huggingface.co/mlx-community](https://huggingface.co/mlx-community/Qwen3-VL-2B-Instruct-4bit/tree/9c4f5209e57b31f4b9dfba735de3fb983739c9cc))
```sh
mlx_lm.chat --model mlx-community/Qwen3-VL-2B-Instruct-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model mlx-community/Qwen3-VL-2B-Instruct-4bit
```
Other versions: 8bit (~2.6 GB, Mac only) `mlx_lm.chat --model lmstudio-community/Qwen3-VL-2B-Instruct-MLX-8bit`
Model repository and documentation: [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct)
[interactive version](https://whatmodel.sh/model/qwen3-vl-2b-instruct) · [JSON](https://whatmodel.sh/api?format=json&all=1)