[whatmodel.sh](../../README.md) / Qwen2 72B Instruct
# Qwen2 72B Instruct
Alibaba · released 2024-06 · 72.71B params · tongyi-qianwen
**score 51.6** ±0.5 · #339 of 732 · from 29 benchmark results across 20 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Qwen2 72B Instruct among every model on that task or test, worst to best.
task vs the others rank counts tests
coding ▁▁▁▃▅▇●▇▅▄▃▁ #263 of 469 17% 4
chat & writing ▁▁▁▃▆▄●█▆▇▇▃ #270 of 402 22% 4
math ▁▁▃▅●▆▄█▄▂▁▁ #293 of 431 17% 6
coding · #263 of 469 · 17% of its score
Beats [Qwen3-VL 32B Instruct](qwen3-vl-32b-instruct.md), [Solar Open 100B (102B-A12B)](solar-open-100b.md), [GLM-4.7-Flash](glm-4-7-flash.md) here, though they rank higher overall. Trails [DeepSeek LLM 67B Chat](deepseek-llm-67b-chat.md), [QwQ-32B-Preview](qwq-32b-preview.md), [InternLM2.5 20B Chat](internlm2-5-20b-chat.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
CRUXEval (ZeroEval, 800 problems) 59.1% ▁▂▆▃●█▆▃▂·▁▁ beats 19/47 medium · avg 57% 6.7% sharp · at its level
LiveCodeBench official (Jan 1–Nov… 25.3% ▁▁·▁▂▄●█▅▂·▂ beats 11/31 hard · avg 23% 5.2% sharp · at its level
Aider code editing (133 Exercism … 55.6% ▁·▁▂▄▁●▄█▆▂▂ beats 24/62 medium · avg 51% 4.9% sharp · at its level
SciCode main problems (official l… 1.5% ▃···●··▂·▃·▂ beats 4/17 very hard · avg 1.0% 0.7% sharp · too hard for it
chat & writing · #270 of 402 · 22% of its score
Beats [Ministral 3 14B Reasoning 2512](ministral-3-14b-reasoning.md), [Ministral 3 14B](ministral-3-14b.md), [OpenThinker2 32B](openthinker2-32b.md) here, though they rank higher overall. Trails [Flan-T5 XXL](flan-t5-xxl.md), [North Micro Vision Instruct (2.4B)](north-micro-vision-instruct.md), [Granite 4.0 H 1B](granite-4-0-h-1b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
AlpacaEval 2 length-controlled 38.1% ▁▂▂▃▆▆█▆●▄▁▁ beats 94/122 medium · avg 35% 11% sharp · at its level
Arena-Hard (v0.1, GPT-4-1106 judg… 46.9% ▂▁▁▅█▅●▃▃▅▃▁ beats 37/60 medium · avg 47% 6.4% sharp · at its level
LMArena (text) 1261.9 ▁▁▁▂▃▃●▆▆▇█▃ beats 98/385 avg 1260 2.6% blunt
Polyrating rating (bias-aware re-… 1182.0 ▂▁▂▁▃▇█▆●▄▂▂ beats 78/95 avg 1187 2.3% typical
math · #293 of 431 · 17% of its score
Beats [Molmo 2 8B](molmo-2-8b.md), [DeepSeek-Coder-V2 Instruct (0724)](deepseek-coder-v2-0724.md), [Llama 3.1 70B Instruct](llama-3-1-70b.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
MATH level 5 39.1% ▂▂▅▆●█▆▃▅▂▃▄ beats 30/92 hard · avg 28% 5.2% sharp · at its level
MATH Level 5 (ZeroEval, 721 probl… 38.3% ▁▁▁▆█▂●▇▄··▂ beats 19/38 hard · avg 30% 3.2% sharp · at its level
Minerva MATH (maker) 31.6% ▃·····▃··●▅▅ beats 4/8 hard · avg 34% 2.7% at its level
GSM8K (ZeroEval, zero-shot) 92.7% ▁▄·▁▄█▅▅▅●██ beats 27/43 very easy · avg 87% 2.6% sharp
GSM8K (maker) 93.2% ▄···█▄▄·●·▄█ beats 5/8 very easy · avg 92% 2.0% sharp · maxed out
MATH (subset unspecified) (maker) 69.0% ▃····▃·●·▅·█ beats 2/8 easy · avg 74% 1.3% typical
other tests (15)
test result vs the others in its run difficulty counts why
LiveBench 2024-07-26 40.2% ▁▁▁▂▂▃▂█▆●▅▃ beats 45/61 medium · avg 38% 11% at its level
MMLU-Redux (ZeroEval, 2,778 quest… 81.6% ▁·▁▁▄▄▄▅●▄▂▁ beats 33/53 easy · avg 75% 10% at its level
ZebraLogic (ZeroEval, 1,000 puzzl… 21.4% ▁▁▃▆●█▁▁·▁▁▁ beats 33/57 hard · avg 15% 3.7% sharp
BigCodeBench (Instruct) (maker) 38.2% ▃▂·▆▆██●▅▃▃▃ beats 21/33 medium · avg 39% 3.1% at its level
GPQA Diamond 40.8% ▁▁▁▁▂●▆▄▅▇█▅ beats 41/211 hard · avg 39% 3.0% sharp
BigCodeBench (Complete) (maker) 53.9% ▂▁▁▄▂▅▇█▆▂●▂ beats 54/58 medium · avg 48% 2.9% at its level
WildBench v2 WB-Score (task macro) 44.4977 WB-Score ▁·▁▁·▃█▄▇●▅▆ beats 33/52 avg 43.019 WB-Score 2.7% sharp
CPTUB average (SpeakLeash, 0-5 ju… 73.1% ▁▂▂▂▃█▃●▅▄▂▃ beats 63/95 medium · avg 65% 1.8% at its level
Polish EQ-Bench v2 (SpeakLeash) 72.1% ▁··▁▁▂▂▆▅▆█● beats 59/70 medium · avg 62% 1.4% at its level
Open Persian LLM Leaderboard v2 a… 65.1% ▂▄▂▃█▄▃▆●▅▂▂ beats 64/81 medium · avg 57% 1.2% blunt · at its level
KINA (Knowledge Index of Noah's A… 24.9% ▁▁▂▁▃●█▆▄▅▂▃ beats 10/39 very hard · avg 21% 1.1% typical
Open PT-LLM Leaderboard average (… 79.1% ▃▁▁▁▁▁▁▁▂█●▁ beats 250/276 medium · avg 72% 0.6% at its level · noisy
RULER 64k 79.8% ▁▁·▁·▂●▄█▂·▁ beats 7/22 easy · avg 82% 0.5% typical
RULER 128k 53.7% ▄··▂▄●▆██▄▄▂ beats 5/21 easy · avg 67% 0.2% at its level · noisy
Vectara hallucination leaderboard… 95.3% ▁▁▂▃▄▄●▆▃▂▂▁ beats 68/140 very easy · avg 94% 0.2% blunt · maxed out · noisy
source results share of the score
WildEval ZeroEval published results 5 27% ▇▇▇
AlpacaEval 1.0 and 2.0 leaderboards 1 11% ▇
LiveBench model judgments 1 11% ▇
Epoch AI 2 8% ▇
Arena-Hard-Auto v0.1 leaderboard 1 6% ▇
Qwen (submitted / maker-reported) 3 6% ▇
BigCodeBench paper (submitted / maker-r… 2 6% ▇
LiveCodeBench official website results 1 5% ▇
Aider code editing benchmark runs 1 5% ▇
WildBench v2 leaderboard data 1 3% ▇
LMArena leaderboard dataset 1 3% ▇
Polyrating: A Cost-Effective and Bias-A… 1 2% ▇
SpeakLeash CPTUB 1 2% ▇
SpeakLeash Polish EQ-Bench leaderboard 1 1% ▇
Open Persian LLM Leaderboard v2 1 1% ▇
Knowledge Index of Noah's Ark 1 1% ▇
SciCode leaderboard 1 1% ▇
NVIDIA RULER repository README results … 2 1% ▇
Open Portuguese LLM Leaderboard 1 1% ▇
Vectara hallucination leaderboard 1 0% ▇
prior: pull toward the middle 0% ▇
weighted average = 51.6
Licences: WildEval ZeroEval published results ([Apache-2.0](../../licenses/zeroeval-published-results-Apache-2.0.txt)) · AlpacaEval 1.0 and 2.0 leaderboards ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0)) · LiveBench model judgments ([Apache-2.0 (stated in the LiveBench datasheet)](https://www.apache.org/licenses/LICENSE-2.0)) · Epoch AI ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Arena-Hard-Auto v0.1 leaderboard ([Apache-2.0](../../licenses/arena-hard-auto-v0-1-Apache-2.0.txt)) · BigCodeBench paper ([CC BY 4.0](http://creativecommons.org/licenses/by/4.0/)) · LiveCodeBench official website results ([CC-BY-SA-4.0](../../licenses/lcb-sharealike-NOTICE.txt)) · Aider code editing benchmark runs ([Apache-2.0](../../licenses/aider-polyglot-runs-Apache-2.0.txt)) · WildBench v2 leaderboard data ([Apache-2.0](https://github.com/allenai/WildBench/blob/d6b8dcaf377d173d031980f97c16e1a82618c03d/LICENSE)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Polyrating: A Cost-Effective and Bias-Aware Rating System for LLM Evaluation ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)) · SpeakLeash CPTUB ([MIT](../../licenses/hf-leaderboard-spaces-MIT.txt)) · SpeakLeash Polish EQ-Bench leaderboard ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt)) · Open Persian LLM Leaderboard v2 ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt)) · Knowledge Index of Noah's Ark ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · SciCode leaderboard ([Apache-2.0](../../licenses/scicode-Apache-2.0.txt)) · NVIDIA RULER repository README results table ([Apache-2.0](https://github.com/NVIDIA/RULER/blob/c3f5e3b4f87f97e048793bb510a3a6b19a46bf3a/LICENSE)) · Open Portuguese LLM Leaderboard ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt)) · Vectara hallucination leaderboard ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0))
## 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 Qwen2 72B Instruct (Q4_K_M, needs ~47.4 GB, from [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen2-72B-Instruct-GGUF/tree/ea5316198ecee4a842976893b2bd2bd2231c5571))
```sh
llama cli -hf Qwen/Qwen2-72B-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/Qwen2-72B-Instruct-GGUF:q4_k_m
```
Other versions: Q8_0 (~77.3 GB) `llama cli -hf Qwen/Qwen2-72B-Instruct-GGUF:q8_0` · Q2_K (~29.8 GB) `llama cli -hf Qwen/Qwen2-72B-Instruct-GGUF:q2_k`
### 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 Qwen2 72B Instruct (Q4_K_M, needs ~47.4 GB, from [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen2-72B-Instruct-GGUF/tree/ea5316198ecee4a842976893b2bd2bd2231c5571))
```sh
lms get https://huggingface.co/Qwen/Qwen2-72B-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 (~77.3 GB) `lms get https://huggingface.co/Qwen/Qwen2-72B-Instruct-GGUF@q8_0 -y` · Q2_K (~29.8 GB) `lms get https://huggingface.co/Qwen/Qwen2-72B-Instruct-GGUF@q2_k -y` · 4bit (~40.9 GB, Mac only) `lms get https://huggingface.co/mlx-community/Qwen2-72B-Instruct-4bit -y` · 8bit (~77.3 GB, Mac only) `lms get https://huggingface.co/mlx-community/Qwen2-72B-Instruct-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 Qwen2 72B Instruct (Q4_0, needs ~44.13 GB, from [ollama.com](https://ollama.com/library/qwen2/tags))
```sh
ollama run qwen2:72b-instruct
```
Other versions: Q8_0 (~87.75 GB) `ollama run qwen2:72b-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 Qwen2 72B Instruct (4bit, needs ~40.9 GB, Mac only, from [huggingface.co/mlx-community](https://huggingface.co/mlx-community/Qwen2-72B-Instruct-4bit/tree/d410561c1c344604ce87755257bc90922b8aa2f6))
```sh
mlx_lm.chat --model mlx-community/Qwen2-72B-Instruct-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model mlx-community/Qwen2-72B-Instruct-4bit
```
Other versions: 8bit (~77.3 GB, Mac only) `mlx_lm.chat --model mlx-community/Qwen2-72B-Instruct-8bit`
Model repository and documentation: [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen2-72B-Instruct)
[interactive version](https://whatmodel.sh/model/qwen2-72b) · [JSON](https://whatmodel.sh/api?format=json&all=1)