[whatmodel.sh](../../README.md) / Qwen2.5 Coder 32B Instruct
# Qwen2.5 Coder 32B Instruct
Alibaba · released 2024-11 · 32.8B params · apache-2.0
**score 57.3** ±0.7 · #239 of 732 · from 25 benchmark results across 11 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Qwen2.5 Coder 32B Instruct among every model on that task or test, worst to best.
task vs the others rank counts tests
coding ▁▁▁▃▅▇█●▅▄▃▁ #188 of 469 87% 17
chat & writing ▁▁▁▃▆▄●█▆▇▇▃ #262 of 402* 4.6% 1
coding · #188 of 469 · 87% of its score
Beats [Qwen3.5 9B](qwen3-5-9b.md), [Gemini 2.5 Flash](gemini-2-5-flash.md), [Devstral 2 (2512)](devstral-2.md) here, though they rank higher overall. Trails [QwQ-32B-Preview](qwq-32b-preview.md), [Ministral 3 8B Reasoning 2512](ministral-3-8b-reasoning.md), [OpenThinker 32B](openthinker-32b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
HumanEval 92.1% ▁·▁▁▃▄▆▅█▃●▁ beats 97/99 easy · avg 76% 17% sharp · several runs
HumanEval+ 87.2% ▁·▁▁▁▃▄▄▄█▃● beats 97/99 easy · avg 70% 17% sharp · several runs
LiveCodeBench official (Jan 1–Nov… 41.6% ▁▁·▁▂▄▆█●▂·▂ beats 25/31 hard · avg 23% 11% sharp · at its level
Aider polyglot (225 exercises, tw… 16.4% ▂·▂●▅▂▃█▆▄▂▂ beats 7/49 very hard · avg 7.6% 10% sharp · at its level
Aider code editing (133 Exercism … 71.4% ▁·▁▂▄▁▆▄█●▂▂ beats 51/62 medium · avg 51% 7.6% sharp · at its level
MBPP+ 77.0% ▁···▁▁▂▅▇█▆● beats 73/75 easy · avg 66% 5.9% several runs
Luau Leetcode fully solved (Torpe… 24.5% ▁▁▁●·▁▂▃▅▃█▇ beats 5/49 hard · avg 17% 3.7% sharp · at its level
WebApp1K-Duo-React pass@1 (onekq-… 41.6% ▁··▁·▁●▁▆█▇█ beats 4/45 hard · avg 29% 2.8% at its level
MBPP (maker) 90.2% █···███·██·● beats 6/6 easy · avg 77% 2.2% typical
MultiPL-E (maker) 79.4% ▄·▄····▄▄·▄● beats 5/6 easy · avg 70% 2.0% typical
WebApp1K-React pass@1 (onekq-ai) 66.8% ▃▁▁█▁▄●▃▄▇▁▄ beats 16/31 medium · avg 40% 1.9% sharp · at its level · noisy
CodeReviewQA: change localisation… 64.3% █▂▁▁▂▂▂▃▅▅▅● beats 43/47 hard · avg 33% 1.7% sharp · at its level · noisy
CodeReviewQA: automated code refi… 44.1% ▁▁▁▁▁▁▂▂▃█▅● beats 41/47 hard · avg 32% 1.3% at its level
CodeReviewQA: solution identifica… 85.7% ▃··▁▂▂▃▂█▃▄● beats 45/47 medium · avg 51% 1.1% sharp · at its level · noisy
CRUXEval-I (maker) 75.2% █·██·█··█·█● beats 6/6 medium · avg 56% 1.1% at its level
CRUXEval-O (maker) 83.4% ▄·█··▄·▄··▄● beats 6/6 medium · avg 54% 0.4% blunt · at its level
CodeReviewQA: change type recogni… 79.1% ▁▁▁······▁▃● beats 46/47 easy · avg 67% 0.0% blunt · at its level · noisy
chat & writing · #262 of 402 · 4.6% of its score
Beats [Mistral Small 4 (119B-A6.5B)](mistral-small-4.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
LMArena (text) 1270.6 ▁▁▁▂▃▃●▆▆▇█▃ beats 102/385 avg 1260 4.6% blunt
other tests (4)
test result vs the others in its run difficulty counts why
DSEffi-Bench Pass (1,000 data-sci… 34.1% ▅··▃●·█·▃██▃ beats 4/15 hard · avg 23% 3.0% at its level
RobloxQA v2.0 multiple choice (To… 84.6% ▁▁●▁█▄▃▃▆▂▂▂ beats 5/49 easy · avg 77% 2.7% typical
BigCodeBench (Instruct) (maker) 49.6% ▄······▄▄█·● beats 6/6 medium · avg 37% 1.6% at its level
Open PT-LLM Leaderboard average (… 74.7% ▃▁▁▁▁▁▁▁▂█●▁ beats 222/276 medium · avg 72% 0.9% at its level · noisy
source results share of the score
EvalPlus / EvalPerf published evaluatio… 3 32% ▇▇▇▇
Qwen (submitted / maker-reported) 8 15% ▇▇
LiveCodeBench official website results 1 11% ▇
Aider polyglot benchmark runs 1 10% ▇
Aider code editing benchmark runs 1 8% ▇
Roblox LLM Leaderboard 2 6% ▇
WebApp1K models leaderboard 2 5% ▇
LMArena leaderboard dataset 1 5% ▇
CodeReviewQA: The Code Review Comprehen… 4 4% ▇
DSEffi-Bench: Demystifying Large Langua… 1 3% ▇
Open Portuguese LLM Leaderboard 1 1% ▇
prior: pull toward the middle 0% ▇
weighted average = 57.3
Licences: EvalPlus / EvalPerf published evaluation runs ([Apache-2.0](../../licenses/evalplus-published-runs-Apache-2.0.txt)) · LiveCodeBench official website results ([CC-BY-SA-4.0](../../licenses/lcb-sharealike-NOTICE.txt)) · Aider polyglot benchmark runs ([Apache-2.0](../../licenses/aider-polyglot-runs-Apache-2.0.txt)) · Aider code editing benchmark runs ([Apache-2.0](../../licenses/aider-polyglot-runs-Apache-2.0.txt)) · Roblox LLM Leaderboard ([MIT](../../licenses/roblox-llm-leaderboard-MIT.txt)) · WebApp1K models leaderboard ([MIT](../../licenses/webapp1k-leaderboard-MIT.txt)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · CodeReviewQA: The Code Review Comprehension Assessment for Large Language Models ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)) · DSEffi-Bench: Demystifying Large Language Models' Capability in Efficient Data Science Code Generation ([CC BY 4.0](../../licenses/paper-2608.30248v1-CC-BY-4.0.txt)) · Open Portuguese LLM Leaderboard ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt))
## How to run it
### Through an API
api (Cloudflare) · $0.74/M blended · $0.66 in / $1 out per 1M tokens · [developers.cloudflare.com](https://developers.cloudflare.com/workers-ai/platform/pricing/)
[price source](https://github.com/cloudflare/cloudflare-docs/blob/b182511e39518f4b3304edd7bb3e0b0d3017b035/src/content/docs/workers-ai/platform/pricing.mdx): Workers AI published marginal USD per 1M tokens after free allocation. Workers plan fees excluded; billed in Neurons. Published USD rates are rounded. Product: Workers AI. API model: @cf/qwen/qwen2.5-coder-32b-instruct. Settings: plan=Workers Paid, free allocation=excluded, billing=published marginal USD rates; service bills Neurons. Currency: USD. · [context source](https://huggingface.co/Qwen/Qwen2.5-Coder-32B-Instruct/blob/381fc969f78efac66bc87ff7ddeadb7e73c218a7/README.md): Full supported model-card context requires the documented YaRN configuration beyond the default 32,768 tokens. This does not promise the same limit on every hosted API.
### 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 Qwen2.5 Coder 32B Instruct (Q4_K_M, needs ~19.9 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Qwen2.5-Coder-32B-Instruct-GGUF/tree/2716de09d455fcc4df73cc202aad2d66c06ac7fc))
```sh
llama cli -hf lmstudio-community/Qwen2.5-Coder-32B-Instruct-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf lmstudio-community/Qwen2.5-Coder-32B-Instruct-GGUF:Q4_K_M
```
Other versions: Q8_0 (~34.8 GB) `llama cli -hf lmstudio-community/Qwen2.5-Coder-32B-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 Qwen2.5 Coder 32B Instruct (Q4_K_M, needs ~19.9 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Qwen2.5-Coder-32B-Instruct-GGUF/tree/2716de09d455fcc4df73cc202aad2d66c06ac7fc))
```sh
lms get https://huggingface.co/lmstudio-community/Qwen2.5-Coder-32B-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 (~34.8 GB) `lms get https://huggingface.co/lmstudio-community/Qwen2.5-Coder-32B-Instruct-GGUF@Q8_0 -y` · 4bit (~18.4 GB, Mac only) `lms get https://huggingface.co/lmstudio-community/Qwen2.5-Coder-32B-Instruct-MLX-4bit -y` · 8bit (~34.8 GB, Mac only) `lms get https://huggingface.co/lmstudio-community/Qwen2.5-Coder-32B-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 Qwen2.5 Coder 32B Instruct (Q4_K_M, needs ~20.18 GB, from [ollama.com](https://ollama.com/library/qwen2.5-coder/tags))
```sh
ollama run qwen2.5-coder:32b
```
Other versions: Q8_0 (~39.86 GB) `ollama run qwen2.5-coder:32b-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 Qwen2.5 Coder 32B Instruct (4bit, needs ~18.4 GB, Mac only, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Qwen2.5-Coder-32B-Instruct-MLX-4bit/tree/96572f69fd88e9c29dd10cd3c6755dd7b30d73d9))
```sh
mlx_lm.chat --model lmstudio-community/Qwen2.5-Coder-32B-Instruct-MLX-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model lmstudio-community/Qwen2.5-Coder-32B-Instruct-MLX-4bit
```
Other versions: 8bit (~34.8 GB, Mac only) `mlx_lm.chat --model lmstudio-community/Qwen2.5-Coder-32B-Instruct-MLX-8bit`
Model repository and documentation: [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen2.5-Coder-32B-Instruct)
[interactive version](https://whatmodel.sh/model/qwen2-5-coder-32b) · [JSON](https://whatmodel.sh/api?format=json&all=1)