whatmodel.sh / Qwen3-VL 8B Thinking

Qwen3-VL 8B Thinking

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

Qwen3-VL 8B Thinking is an open-weights language model from Alibaba, released October 2025. Its published size is 8.77B parameters and a 256k-token context window. Licence: apache-2.0.

Catalogue identity notes

Maker repo Qwen/Qwen3-VL-8B-Thinking @92f3c4b (licence tag apache-2.0, created 2025-10-11); the separate Thinking checkpoint, not Qwen3-VL-8B-Instruct. Scored from the Qwen3-VL Technical Report (arXiv 2511.21631v2, Tables 3/4 and 7-10: column 'Qwen3-VL 8B thinking'), whose instruct and thinking columns are the two published checkpoints. Released 2025-10: month the maker's Hugging Face repo was created.

Model information sources

Release date ↗ · Weights licence ↗

8,767,123,696 parameters in the Hugging Face safetensors metadata of Qwen/Qwen3-VL-8B-Thinking @92f3c4b (vision encoder included); dense, so active = total.

57.8±0.7#222 of 719 LLMsfrom 28 benchmark results across 3 sources

$ how it compares

The LLMs ranked either side of Qwen3-VL 8B Thinking, out of 828 with a score, then more from Alibaba.

#modelscorewhereRAMreleased
264Nemotron-Cascade-8B-Thinking NVIDIA58±2local~6.3GB2025-12
265Qwen3-VL 30B-A3B Instruct Alibaba58±1api $0.35/M +local~21GB2025-09
266Qwen3-VL 8B Thinking Alibaba58±1api $0.66/M +local~6.8GB2025-10
267Qwen3 32B Alibaba58±0api $0.13/M +local~20GB2025-04
268Qwen3 Coder 30B-A3B Instruct Alibaba58±1api $0.12/M +local~20GB2025-07
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

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #201 of 300 * · 1.9% of its score · 1 test

Beats o1 (2024-12-17), o3-mini, DeepSeek R1 (Jan '25) here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Mini, Mellum2 12B-A2.5B Instruct here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
BFCL v3 lab63.0% ✎█▄··▄█·█▄█▄█ beats 4/13medium · avg 59%▇ 1.9%blunt · at its level
coding▁▁▁▃▆██▇▆▄▃▁ #210 of 458 * · 4.2% of its score · 1 test

Beats Qwen3.5 4B, Qwen3-VL 32B Instruct, Solar Open 100B (102B-A12B) here, though they rank higher overall. Trails QwQ-32B-Preview, Ministral 3 3B Reasoning 2512, DeepSeek-Coder-V2 Instruct (0724) here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
LiveCodeBench v6 (window unspecified) lab58.6% ✎▃▅█▅▅▃▅···▃█ beats 10/16medium · avg 35%▇▇ 4.2%at its level
chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #196 of 393 * · 2.5% of its score · 1 test

Beats Granite 4.2 8B, Qwen3-VL 30B-A3B Thinking, Mistral Small 4 (119B-A6.5B) here, though they rank higher overall. Trails LFM2.5 1.2B Instruct, LFM2.5 1.2B Thinking, Gemma 3 12B here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
IFEval (aggregation unspecified) lab83.2% ✎▃··▃···██▅▃▃ beats 5/11easy · avg 77%▇ 2.5%typical
math▁▂▃▅▆▅▄█▄▁▁▁ #147 of 398 · 16% of its score · 4 tests

Beats Claude Sonnet 4.6, MiMo-V2.5 (310B-A15B), MiniMax-M2.7 here, though they rank higher overall. Trails Nemotron 3 Nano 30B-A3B here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
MathVista (testmini) lab81.4% ✎▃··▃▅·▃█▅▅▅█ beats 12/16medium · avg 69%▇▇ 4.9%typical
AIME 2025 lab80.3% ✎▃·▃▃█··▃▅▃·█ beats 10/12medium · avg 45%▇▇ 4.2%sharp · at its level
OMEGA (math generalisation)38.1% ✎▄······█▄▄·█ beats 2/6hard · avg 30%▇▇ 4.0%at its level
MATH (subset unspecified)95.2% ✎▃·····▃···▅█ beats 6/6very easy · avg 90%▇ 2.9%maxed out
vision & documents▁▁▃▂▁▃▄▂▆▇█▄ #94 of 147 · 47% of its score · 14 tests

Beats DeepSeek V4 Flash Vision (Exp), Magistral Small 1.2, Qwen3-VL 30B-A3B Instruct here, though they rank higher overall. Trails GPT-4o (2024-11-20) here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
CharXiv (reasoning questions) lab53.0% ✎▄▂▄▆▄▄▆··▄█▂ beats 9/21medium · avg 39%▇▇▇▇ 8.9%at its level
MMMU (val) lab74.1% ✎▂···▆▂▂▂▄▄▄█ beats 11/16medium · avg 61%▇▇▇ 7.3%at its level
CharXiv descriptive questions lab85.9% ✎▄··▂▂▂·▂▄▄▂█ beats 8/14easy · avg 74%▇▇ 5.5%typical
MathVista (testmini) lab81.4% ✎▃··▃▅·▃█▅▅▅█ beats 12/16medium · avg 69%▇▇ 4.9%typical
MMStar lab75.3% ✎▂··▄▂▂·▄█▂▄▆ beats 12/16medium · avg 65%▇▇ 4.8%at its level
MMMU Pro lab60.4% ✎▃·█·▅▃▃▃▅▅▃█ beats 11/16medium · avg 46%▇▇ 4.3%at its level
ChartArena chart parsing (English, mAP_high average)39.6% ✎▁·▁·▃▃▁▇▃█·▄ beats 12/22hard · avg 27%▇ 2.5%blunt · at its level
HallusionBench lab65.4% ✎▃·▃·▅▃▅▅▃▅█▅ beats 14/16medium · avg 55%▇ 2.0%blunt · at its level
6 more tests
testresultvs the others in its rundifficultycountswhy
ChartQA lab88.6% ✎▂··▂·▂▃·▅▂▆█ beats 10/17easy · avg 85%▇ 1.4%blunt
RealWorldQA lab73.5% ✎▅▃·▃▅▅·█·▃▅█ beats 9/16easy · avg 70%▇ 1.3%blunt
DocVQA lab95.3% ✎▃··▅▃▅▃█▃█▅▅ beats 7/17very easy · avg 94%▇ 1.2%blunt · maxed out
ScreenSpot-Pro lab46.6% ✎▂····▃▄··▂█▆ beats 1/11medium · avg 46%▇ 1.2%blunt · at its level
CC-OCR lab76.3% ✎▃··▃▅·▅▅·█▅▃ beats 5/13easy · avg 74%▇ 0.9%blunt
OCRBench lab81.9% ✎▅█·▃▃█▃▅▅▃·▃ beats 4/16easy · avg 84%▇ 0.5%blunt
other tests8 not in a task · 33% of its score
testresultvs the others in its rundifficultycountswhy
GPQA (subset unspecified) lab69.9% ✎▂····▂·▂▂█▆█ beats 9/14hard · avg 49%▇▇▇ 8.4%sharp · at its level
MMLU-Pro lab77.3% ✎▄·▄·▄▄▄▄█▄██ beats 8/12medium · avg 63%▇▇▇ 6.3%at its level
MMLU-Redux lab88.8% ✎▃··▃·▃·█▃▅▃▅ beats 8/11easy · avg 79%▇▇ 4.9%typical
SuperGPQA lab51.2% ✎▄··▄·▄▄▄█▄██ beats 7/11medium · avg 37%▇▇ 4.2%at its level
AndroidWorld lab50.0% ✎▃··▃▅▃▃▃▃▃·█ beats 4/11medium · avg 43%▇ 3.0%blunt · at its level
LVBench lab55.8% ✎▅··▃·▅▃▃·█·▃ beats 3/10hard · avg 51%▇ 2.9%at its level
Video-MME (without subtitles) lab71.8% ✎▃··█··▅█▃▃▅▃ beats 8/13medium · avg 66%▇ 2.6%blunt · at its level
MVBench lab69.0% ✎▂▂▂·▂█··▆▄·▂ beats 6/13medium · avg 67%▇ 1.1%blunt · at its level

$ where to run it

Through an API

api · Alibaba Cloud Model Studio · $0.66/M · $0.18 in / $2.1 out per 1M tokens · 256k context · www.alibabacloud.com ↗ · price sourceUSD per 1M tokens, Alibaba Cloud Model Studio, Singapore (International deployment scope), API model qwen3-vl-8b-thinking (Thinking mode only): input $0.18, output $2.1. Chinese-mainland/Global (US) deployments priced differently and not used, following the International convention of other Alibaba rows; free quota, batch and context-cache discounts excluded. Observed 2026-10-09.

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 Qwen3-VL 8B Thinking · needs ~5 GB · from huggingface.co/Qwen (official)
    chat in the terminal
    llama cli -hf Qwen/Qwen3-VL-8B-Thinking-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf Qwen/Qwen3-VL-8B-Thinking-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 Qwen3-VL 8B Thinking · needs ~5 GB · from huggingface.co/Qwen (official)

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/Qwen/Qwen3-VL-8B-Thinking-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 Qwen3-VL 8B Thinking · needs ~6.1 GB · from ollama.com
    chat in the terminal
    ollama run qwen3-vl:8b-thinking

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 Qwen3-VL 8B Thinking · needs ~5.8 GB, Mac only · from huggingface.co/lmstudio-community
    chat in the terminal
    mlx_lm.chat --model lmstudio-community/Qwen3-VL-8B-Thinking-MLX-4bit
  3. Or start an OpenAI-compatible server at http://localhost:8080/v1
    mlx_lm.server --model lmstudio-community/Qwen3-VL-8B-Thinking-MLX-4bit

Official weights and documentation: huggingface.co/Qwen ↗

$ how the score was worked out · Qwen (Alibaba) vendor-reporte… 90%, allenai/Olmo-3-7B-Think model… 7%, ChartArena: Benchmarking Char… 2%

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-reported25▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 90%Factual vendor-reported score (not licensed content)
allenai/Olmo-3-7B-Think model card2▇▇ 7%Apache-2.0
ChartArena: Benchmarking Chart Parsing across Languages, Scenarios, and Formats (arXiv:2606.01348v3)1▇ 2%CC BY 4.0
prior · pull toward the middle▇ 0%
weighted average= 57.8

$ 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