[whatmodel.sh](../../README.md) / Qwen3-VL 8B Thinking
# Qwen3-VL 8B Thinking
Alibaba · released 2025-10 · 8.77B params · apache-2.0
**score 57.5** ±0.7 · #236 of 732 · from 28 benchmark results across 3 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 8B Thinking among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃▆●▄▅▅▄▄▂▁ #213 of 313* 2.0% 1
coding ▁▁▁▃▅▇●▇▅▄▃▁ #225 of 469* 4.0% 1
chat & writing ▁▁▁▃▆▄▆●▆▇▇▃ #204 of 402* 2.6% 1
math ▁▁▃▅█▆▄●▄▂▁▁ #149 of 431 16% 4
vision & documents ▁▁▂▂▁▃●▂▆▇█▄ #100 of 153 47% 14
agent tasks · #213 of 313 · 2.0% of its score
Beats [o1 (2024-12-17)](o1-2024-12-17.md), [o3-mini](o3-mini.md), [DeepSeek R1 (Jan '25)](deepseek-r1.md) here, though they rank higher overall. Trails [SmolLM3 3B](smollm3-3b.md), [Jamba 1.5 Mini](jamba-1-5-mini.md), [Mellum2 12B-A2.5B Instruct](mellum2-12b-a2-5b-instruct.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
BFCL v3 (maker) 63.0% █▄··▄●·█▄█▄█ beats 4/13 medium · avg 59% 2.0% blunt · at its level
coding · #225 of 469 · 4.0% of its score
Beats [Qwen3.5 4B](qwen3-5-4b.md), [Qwen3-VL 32B Instruct](qwen3-vl-32b-instruct.md), [Solar Open 100B (102B-A12B)](solar-open-100b.md) here, though they rank higher overall. Trails [QwQ-32B-Preview](qwq-32b-preview.md), [Ministral 3 3B Reasoning 2512](ministral-3-3b-reasoning.md), [s1.1 32B](s1-1-32b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LiveCodeBench v6 (window unspecif… 58.6% ▃▅█▅▅●▅···▃█ beats 10/16 hard · avg 35% 4.0% at its level
chat & writing · #204 of 402 · 2.6% of its score
Beats [Gemma 2 9B IT SimPO](gemma-2-9b-simpo.md), [Granite 4.2 8B](granite-4-2-8b.md), [Mistral Small 4 (119B-A6.5B)](mistral-small-4.md) here, though they rank higher overall. Trails [LFM2.5 1.2B Instruct](lfm2-5-1-2b-instruct.md), [LFM2.5 1.2B Thinking](lfm2-5-1-2b-thinking.md), [LFM2.5 8B-A1B](lfm2-5-8b-a1b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
IFEval (aggregation unspecified) … 83.2% ▃··▃···█●▅▃▃ beats 5/11 easy · avg 77% 2.6% typical
math · #149 of 431 · 16% of its score
Beats [Claude Sonnet 4.6](claude-sonnet-4-6.md), [MiMo-V2.5 (310B-A15B)](mimo-v2-5.md), [MiniMax-M2.7](minimax-m2-7.md) here, though they rank higher overall.
test result vs the others in its run difficulty counts why
MathVista (testmini) (maker) 81.4% ▃··▃▅·▃█▅▅●█ beats 12/16 medium · avg 69% 4.7% at its level
AIME 2025 (maker) 80.3% ▃·▃▃█··▃▅▃·● beats 10/12 medium · avg 44% 4.3% sharp · at its level
OMEGA (math generalisation) 38.1% ▄······●▄▄·█ beats 2/6 hard · avg 31% 3.6% at its level
MATH (subset unspecified) 95.2% ▃·····▃···▅● beats 6/6 very easy · avg 91% 3.0% maxed out
vision & documents · #100 of 153 · 47% of its score
Beats [DeepSeek V4 Flash Vision (Exp)](deepseek-v4-flash-vision-exp.md), [GPT-4o (2024-08-06)](gpt-4o-2024-08-06.md), [GPT-4o (2024-05-13)](gpt-4o-2024-05-13.md) here, though they rank higher overall.
test result vs the others in its run difficulty counts why
CharXiv (reasoning questions) (ma… 53.0% ▄▂▄▆●▄▆··▄█▂ beats 9/21 medium · avg 38% 9.3% at its level
MMMU (val) (maker) 74.1% ▂···▆▂▂▂▄▄●█ beats 11/16 medium · avg 61% 7.1% at its level
CharXiv descriptive questions (ma… 85.9% ▄··▂▂▂·▂▄●▂█ beats 8/14 easy · avg 74% 5.7% typical
MMStar (maker) 75.3% ▂··▄▂▂·▄█▂●▆ beats 12/16 medium · avg 65% 4.8% at its level
MathVista (testmini) (maker) 81.4% ▃··▃▅·▃█▅▅●█ beats 12/16 medium · avg 69% 4.7% at its level
MMMU Pro (maker) 60.4% ▃·█·▅▃▃▃▅●▃█ beats 11/16 medium · avg 46% 4.3% at its level
ChartArena chart parsing (English… 39.6% ▁·▁·▃▃▁▇●█·▄ beats 12/22 hard · avg 27% 2.6% blunt · at its level
HallusionBench (maker) 65.4% ▃·▃·▅▃▅▅▃▅●▅ beats 14/16 medium · avg 54% 2.0% blunt · at its level
ChartQA (maker) 88.6% ▂··▂·▂▃·▅▂●█ beats 10/17 easy · avg 85% 1.4% blunt
RealWorldQA (maker) 73.5% ▅▃·▃▅▅·●·▃▅█ beats 9/16 easy · avg 70% 1.4% blunt
DocVQA (maker) 95.3% ▃··▅▃▅▃●▃█▅▅ beats 7/17 very easy · avg 94% 1.2% blunt · maxed out
ScreenSpot-Pro (maker) 46.6% ▂····●▄··▂█▆ beats 1/11 medium · avg 46% 1.2% blunt · at its level
CC-OCR (maker) 76.3% ▃··▃▅·●▅·█▅▃ beats 5/13 easy · avg 74% 0.9% blunt
OCRBench (maker) 81.9% ▅●·▃▃█▃▅▅▃·▃ beats 4/16 easy · avg 84% 0.5% blunt
other tests (8)
test result vs the others in its run difficulty counts why
GPQA (subset unspecified) (maker) 69.9% ▂····▂·▂▂█●█ beats 9/14 hard · avg 49% 8.3% sharp · at its level
MMLU-Pro (maker) 77.3% ▄·▄·▄▄▄▄█●██ beats 8/12 medium · avg 63% 6.4% at its level
MMLU-Redux (maker) 88.8% ▃··▃·▃·█▃●▃▅ beats 8/11 easy · avg 79% 4.8% typical
SuperGPQA (maker) 51.2% ▄··▄·▄▄▄█●██ beats 7/11 medium · avg 37% 4.0% at its level
AndroidWorld (maker) 50.0% ▃··▃▅●▃▃▃▃·█ beats 4/11 medium · avg 43% 3.0% at its level
LVBench (maker) 55.8% ▅··▃·●▃▃·█·▃ beats 3/10 hard · avg 51% 2.9% at its level
Video-MME (without subtitles) (ma… 71.8% ▃··█··▅●▃▃▅▃ beats 8/13 medium · avg 66% 2.7% blunt · at its level
MVBench (maker) 69.0% ▂▂▂·▂●··▆▄·▂ beats 6/13 medium · avg 67% 1.1% blunt · at its level
source results share of the score
Qwen (submitted / maker-reported) 25 91% ▇▇▇▇▇▇▇▇▇▇▇
allenai/Olmo-3-7B-Think model card 2 7% ▇
ChartArena: Benchmarking Chart Parsing … 1 3% ▇
prior: pull toward the middle 0% ▇
weighted average = 57.5
Licences: allenai/Olmo-3-7B-Think model card ([Apache-2.0](../../licenses/allenai--Olmo-3-7B-Think.txt)) · ChartArena: Benchmarking Chart Parsing across Languages ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/))
## How to run it
### Through an API
api (Alibaba Cloud Model Studio) · $0.66/M blended · $0.18 in / $2.1 out per 1M tokens · [alibabacloud.com](https://www.alibabacloud.com/help/en/model-studio/model-pricing)
[price source](https://www.alibabacloud.com/help/en/model-studio/model-pricing): USD 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. · [context source](https://huggingface.co/Qwen/Qwen3-VL-8B-Thinking/blob/92f3c4b4feadd3a016ef468d103bb5f58b2a2c6b/README.md): Official Qwen model card: 'Native 256K context, expandable to 1M'. Observed 2026-10-09.
### 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 8B Thinking (Q4_K_M, needs ~5 GB, from [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen3-VL-8B-Thinking-GGUF/tree/bca5838231f8cc1303cf8810afffcfbdc41bc75a))
```sh
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)
```sh
llama serve -hf Qwen/Qwen3-VL-8B-Thinking-GGUF:Q4_K_M
```
Other versions: Q8_0 (~8.7 GB) `llama cli -hf Qwen/Qwen3-VL-8B-Thinking-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 8B Thinking (Q4_K_M, needs ~5 GB, from [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen3-VL-8B-Thinking-GGUF/tree/bca5838231f8cc1303cf8810afffcfbdc41bc75a))
```sh
lms get https://huggingface.co/Qwen/Qwen3-VL-8B-Thinking-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 (~8.7 GB) `lms get https://huggingface.co/Qwen/Qwen3-VL-8B-Thinking-GGUF@Q8_0 -y` · 4bit (~5.8 GB, Mac only) `lms get https://huggingface.co/lmstudio-community/Qwen3-VL-8B-Thinking-MLX-4bit -y` · 8bit (~9.9 GB, Mac only) `lms get https://huggingface.co/lmstudio-community/Qwen3-VL-8B-Thinking-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 8B Thinking (Q4_K_M, needs ~6.1 GB, from [ollama.com](https://ollama.com/library/qwen3-vl/tags))
```sh
ollama run qwen3-vl:8b-thinking
```
Other versions: Q8_0 (~9.8 GB) `ollama run qwen3-vl:8b-thinking-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 8B Thinking (4bit, needs ~5.8 GB, Mac only, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Qwen3-VL-8B-Thinking-MLX-4bit/tree/51f2ebed945e9124130b5dafe33df0ceb2418833))
```sh
mlx_lm.chat --model lmstudio-community/Qwen3-VL-8B-Thinking-MLX-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model lmstudio-community/Qwen3-VL-8B-Thinking-MLX-4bit
```
Other versions: 8bit (~9.9 GB, Mac only) `mlx_lm.chat --model lmstudio-community/Qwen3-VL-8B-Thinking-MLX-8bit`
Model repository and documentation: [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen3-VL-8B-Thinking)
[interactive version](https://whatmodel.sh/model/qwen3-vl-8b-thinking) · [JSON](https://whatmodel.sh/api?format=json&all=1)