whatmodel.sh / Qwen2.5-VL 3B Instruct

Qwen2.5-VL 3B Instruct

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Qwen2.5-VL 3B Instruct is an open-weights language model from Alibaba, released January 2025. Its published size is 3.75B parameters. Licence: qwen-research.

Catalogue identity notes

Qwen's Qwen2.5-VL 3B Instruct (Qwen/Qwen2.5-VL-3B-Instruct @6628554, created 2025-01-26, licence qwen-research). The Qwen2.5-VL technical report (arXiv 2502.13923v1) evaluates the released Instruct models; its 'Qwen2.5-VL 3B' columns are this checkpoint. ollama qwen2.5vl:3b tags package the same model. Smaller siblings of the catalogued qwen2-5-vl-32b / qwen2-5-vl-72b.

Model information sources

Release date ↗ · Weights licence ↗

HF safetensors metadata (dense).

44.1±1.0#458 of 719 LLMsfrom 14 benchmark results across 2 sources

$ how it compares

The LLMs ranked either side of Qwen2.5-VL 3B Instruct, out of 828 with a score, then more from Alibaba.

#modelscorewhereRAMreleased
545Eurus-2-7B-PRIME PRIME-RL44±2local~6.4GB2024-12
546Llama 3.1 Tulu 3 8B Ai244±1local~6.2GB2024-11
547Qwen2.5-VL 3B Instruct Alibaba44±1local~3.5GB2025-01
548Qwen1.5 32B Chat Alibaba44±1local~22GB2024-04
549Granite 20B Code Instruct (8K) IBM44±1local~14GB2024-04
more from Alibaba
18Qwen3.8 Max (2.4T-A95B) Alibaba78±1api $3.00/M +local~380GB2026-08
34Qwen3.8 Max (0902) Alibaba76±1api $3.00/M2026-09
37Qwen3.8-Flash-Next Alibaba75±1local~59GB2026-08

see all LLMs · find the ones that fit your machine

$ what it's good at

math▁▂▃▅▆▅▄█▄▁▁▁ #288 of 398 * · 13% of its score · 1 test

Beats GPT-4 (0613), Athene-70B, Gemma 2 27B IT here, though they rank higher overall. Trails DeepSeek R1 Distill Qwen 1.5B, Phi-2 (2.7B), StarCoder2 15B here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
MathVista (testmini) lab62.3% ✎▃··▃▅·▃█▅▅▅█ beats 3/16medium · avg 69%▇▇▇▇▇ 13%at its level
vision & documents▁▁▃▂▁▃▄▂▆▇█▄ #133 of 147 · 100% of its score · 14 tests

Beats LLaVA-OneVision Qwen2 72B (ov-sft), Gemma 3 4B here, though they rank higher overall. Trails Qwen3-VL 2B Instruct here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
CharXiv descriptive questions lab58.6% ✎▄··▂▂▂·▂▄▄▂█ beats 0/14easy · avg 74%▇▇▇▇▇▇▇ 18%at its level
MMMU (val) lab53.1% ✎▂···▆▂▂▂▄▄▄█ beats 1/16medium · avg 61%▇▇▇▇▇▇ 15%at its level
CharXiv (reasoning questions) lab31.3% ✎▄▂▄▆▄▄▆··▄█▂ beats 1/21medium · avg 39%▇▇▇▇▇▇ 14%at its level
MathVista (testmini) lab62.3% ✎▃··▃▅·▃█▅▅▅█ beats 3/16medium · avg 69%▇▇▇▇▇ 13%at its level
MMStar lab55.9% ✎▂··▄▂▂·▄█▂▄▆ beats 1/16medium · avg 65%▇▇▇▇ 10%at its level
MMMU Pro lab31.6% ✎▃·█·▅▃▃▃▅▅▃█ beats 0/16medium · avg 46%▇▇▇ 6.9%at its level
HallusionBench lab46.3% ✎▃·▃·▅▃▅▅▃▅█▅ beats 1/16medium · avg 55%▇▇ 4.1%blunt · at its level
ChartQA lab84.0% ✎▂··▂·▂▃·▅▂▆█ beats 3/17easy · avg 85%▇▇ 3.9%blunt
6 more tests
testresultvs the others in its rundifficultycountswhy
DocVQA lab93.9% ✎▃··▅▃▅▃█▃█▅▅ beats 3/17very easy · avg 94%▇▇ 3.9%blunt · maxed out
ChartArena chart parsing (English, mAP_high average)21.4% ✎▁·▁·▃▃▁▇▃█·▄ beats 2/22hard · avg 27%▇ 3.5%blunt
RealWorldQA lab65.4% ✎▅▃·▃▅▅·█·▃▅█ beats 2/16easy · avg 70%▇ 3.2%blunt · at its level
CC-OCR lab74.5% ✎▃··▃▅·▅▅·█▅▃ beats 3/13easy · avg 74%▇ 2.2%blunt · at its level
OCRBench lab79.7% ✎▅█·▃▃█▃▅▅▃·▃ beats 1/16easy · avg 84%▇ 1.2%blunt
TextVQA lab79.3% ✎█······▄·▄▄▄ beats 0/5easy · avg 80%▇ 0.7%blunt

$ where to run it

llama.cpp · recommended

version

  1. 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 ↗

  2. Run Qwen2.5-VL 3B Instruct · needs ~1.9 GB · from huggingface.co/ggml-org
    chat in the terminal
    llama cli -hf ggml-org/Qwen2.5-VL-3B-Instruct-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf ggml-org/Qwen2.5-VL-3B-Instruct-GGUF:Q4_K_M

LM Studio

version

  1. Install LM Studio · Open the app once before using the lms command.

    download it (macOS / Windows / Linux) ↗

  2. Run Qwen2.5-VL 3B Instruct · needs ~1.9 GB · from huggingface.co/ggml-org

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/ggml-org/Qwen2.5-VL-3B-Instruct-GGUF@Q4_K_M -y
  4. Then run lms ls to see its name, and lms chat followed by that name

Ollama

version

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run Qwen2.5-VL 3B Instruct · needs ~3.2 GB · from ollama.com
    chat in the terminal
    ollama run qwen2.5vl:3b-q4_K_M

MLX (Mac)

version

  1. Install MLX (Mac) · Apple silicon Macs only, text only.
    with uv
    uv tool install mlx-lm
    or
    pip install mlx-lm
  2. Run Qwen2.5-VL 3B Instruct · needs ~3.1 GB, Mac only · from huggingface.co/mlx-community
    chat in the terminal
    mlx_lm.chat --model mlx-community/Qwen2.5-VL-3B-Instruct-4bit
  3. Or start an OpenAI-compatible server at http://localhost:8080/v1
    mlx_lm.server --model mlx-community/Qwen2.5-VL-3B-Instruct-4bit

Official weights and documentation: huggingface.co/Qwen ↗

$ how the score was worked out · Qwen (Alibaba) vendor-reporte… 96%, ChartArena: Benchmarking Char… 4%

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

sourceresultsshare of the scorelicence
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-reported13▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 96%Factual vendor-reported score (not licensed content)
ChartArena: Benchmarking Chart Parsing across Languages, Scenarios, and Formats (arXiv:2606.01348v3)1▇ 4%CC BY 4.0
prior · pull toward the middle▇ 0%
weighted average= 44.1

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit 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 · sources and licences · json