AlibabaQwen 2.5Qwen2ForCausalLM

Qwen2.5 Coder 32B — Hardware Requirements & GPU Compatibility

ChatCodeFunctions

Qwen2.5-Coder-32B is Alibaba's 32.8-billion-parameter code-specialized base language model, the largest of six sizes in the Qwen2.5-Coder family (0.5B to 32B), continued-pretrained from Qwen2.5-32B on 5.5 trillion tokens of source code, text-code grounding data, and synthetic data. It is a raw pretrained model, not instruction-tuned, so it is meant as a foundation for building code assistants, agents, or chat models through further fine-tuning rather than for direct conversational use. Alibaba states its coding ability rivals GPT-4o, and at this size it needs a high-end consumer GPU or multi-GPU setup once quantized, more at full precision. Context length defaults to 32,768 tokens, extendable to 131,072 tokens with YaRN scaling. It is released under the Apache 2.0 license, permitting unrestricted commercial and research use, and was published in November 2024.

32.8B params · dense33K contextApache 2.0Nov 20242.2K downloads190 likes15.8K quant downloads

Based on Qwen2.5 32B

Your hardware

Verdicts below are for the reference RTX 4090. Set yours to personalise the page.

Detection reads your browser's GPU name with no data sent anywhere.

Runs well on a 24 GB GPU

Runs from 16 GB (IQ2_M · 16K)Best on 24 GB (Q4_K_M · ~32 tok/s · 8K)Macs from 18 GB (IQ2_XS · ~10 tok/s)

Apache 2.0

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Qwen2.5 Coder 32B

One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.

Ways to runOther ways

each opens step-by-step

Run in cloud

Your own private copy · Q4_K_M · 32 GB GPU

The easiest way to try this model
Running in a few minutes, easy setup
Chat right here or connect any app

$5+ bonus for new users

Run Qwen2.5 Coder 32B

24.3 GB VRAM at 16K context

Works on macOS, Windows and Linux.

Install Ollama (Linux)
$curl -fsSL https://ollama.com/install.sh | sh
Install Ollama (macOS / Windows)
Download and run the app; it installs the ollama command.
Download Ollama ↗
Run Q4_K_M
$ollama run hf.co/lmstudio-community/Qwen2.5-Coder-32B-GGUF:Q4_K_M
Raise the context window
Type it inside the chat. For the API server instead, start it with OLLAMA_CONTEXT_LENGTH=16384 ollama serve.
$/set parameter num_ctx 16384
A larger context needs more memory than the weights alone.
Use the API
Ollama serves its API on http://localhost:11434.

Works on macOS, Windows and Linux.

Install LM Studio
Desktop app for macOS (Apple silicon), Windows and Linux.
lmstudio.ai ↗
Download the model
In the app press Ctrl/Cmd + Shift + M and paste https://huggingface.co/lmstudio-community/Qwen2.5-Coder-32B-GGUF, then pick Q4_K_M. Or use the CLI (run LM Studio once first):
$lms get https://huggingface.co/lmstudio-community/Qwen2.5-Coder-32B-GGUF@Q4_K_M
Open on Hugging Face ↗
Load and chat
Open the model loader, select the model and set Context Length to 16384. Leave GPU offload at maximum if it fits.
Start the local server (optional)
$lms server start

Works on macOS, Windows and Linux.

Install llama.cpp (macOS)
MacPorts, Nix and conda-forge also work.
$brew install llama.cpp
Install options ↗
Install llama.cpp (Windows)
$winget install llama.cpp
Install llama.cpp (Linux)
Download the prebuilt llama-*-bin-ubuntu-* archive for your GPU backend (CPU, Vulkan) from the releases page, unpack it and run the binaries from that folder. Homebrew (brew install llama.cpp) and conda-forge also work on Linux.
llama.cpp releases ↗
Start the server
Downloads the model on first run. The chat UI and an OpenAI-compatible API are at http://localhost:8080.
$llama-server -hf lmstudio-community/Qwen2.5-Coder-32B-GGUF:Q4_K_M -c 16384 -ngl 99 --port 8080

Works on macOS, Windows and Linux.

Option A: Docker Model Runner
Docker Desktop: Settings > AI > Enable Docker Model Runner. Docker Engine on Linux: install the docker-model-plugin package.
$docker model run hf.co/lmstudio-community/Qwen2.5-Coder-32B-GGUF:Q4_K_M
Model Runner docs ↗
The quant tag after the repo is how Hugging Face cards write it; if your Docker version rejects it, drop the tag and pick the quant by repo.
Option B: llama.cpp server image (NVIDIA)
Needs the NVIDIA Container Toolkit. Models are cached in the llama-cache volume. Open http://localhost:8080.
$docker run --gpus all -p 8080:8080 -v llama-cache:/root/.cache ghcr.io/ggml-org/llama.cpp:server-cuda -hf lmstudio-community/Qwen2.5-Coder-32B-GGUF:Q4_K_M -c 16384 -ngl 99 --host 0.0.0.0 --port 8080
Image tags (cuda13, vulkan, rocm) ↗
Cache path inside the image is not in the docs; remove the -v flag if the volume complains.

Linux only.

Install vLLM
Linux only; on Windows run it inside WSL2. NVIDIA GPU shown; AMD ROCm, Intel and CPU builds have their own install pages.
$uv venv --python 3.12 --seed
$source .venv/bin/activate
$uv pip install vllm --torch-backend=auto
vLLM installation ↗
Serve the model
OpenAI-compatible API at http://localhost:8000/v1.
$vllm serve Qwen/Qwen2.5-Coder-32B --max-model-len 16384
Runs the original weights, not the Q4_K_M GGUF; needs ~72 GB of GPU memory.

macOS only.

MLX supports only macOS, so follow these steps for that setup.

Install mlx-lm
$pip install mlx-lm
Chat in the terminal
$mlx_lm.chat --model mlx-community/Qwen2.5-Coder-32B-4bit
Downloads the model on first run. The MLX repo has its own quantisation; it may not match the quant selected above.
Or start a local server
Listens on http://localhost:8080.
$mlx_lm.server --model mlx-community/Qwen2.5-Coder-32B-4bit

Works on macOS, Windows and Linux.

Install Jan
jan.ai ↗
Find the model
Open the Hub tab, paste this model ID in Search, and choose Q4_K_M if offered.
$lmstudio-community/Qwen2.5-Coder-32B-GGUF
Use the local API (optional)
Settings > Local API Server > Start Server. It listens on http://127.0.0.1:1337.

Rent GPUs hourly to run Q4_K_M right there. Pricing comes from partner marketplaces and varies.

Available GPU rentals for Q4_K_M, prioritized by value
GPU	Speed	$/1M tokens	Rent
RTX 5090 (32 GB)CheapestCheapest7 GB free	~57 tok/s	$2.43	Deploy on Vast.aifrom $0.50/hr (opens in new tab)	32 GB · 7 GB free · ~57 tok/s · $2.43 / 1M tokens
RTX PRO 4500 (32 GB)7 GB free	~28 tok/s	$3.33	Deploy on RunPodfrom $0.34/hr (opens in new tab)	32 GB · 7 GB free · ~28 tok/s · $3.33 / 1M tokens
A40 (48 GB)23 GB free	~22 tok/s	$4.40	Deploy on RunPodfrom $0.35/hr (opens in new tab)	48 GB · 23 GB free · ~22 tok/s · $4.40 / 1M tokens
RTX PRO 4500 (32 GB)7 GB free	~28 tok/s	$3.48	Deploy on Vast.aifrom $0.36/hr (opens in new tab)	32 GB · 7 GB free · ~28 tok/s · $3.48 / 1M tokens
RTX A6000 (48 GB)23 GB free	~24 tok/s	$4.60	Deploy on Vast.aifrom $0.40/hr (opens in new tab)	48 GB · 23 GB free · ~24 tok/s · $4.60 / 1M tokens
A100 40GB15 GB free	~49 tok/s	$3.44	Deploy on Vast.aifrom $0.61/hr (opens in new tab)	15 GB free · ~49 tok/s · $3.44 / 1M tokens
L40 (48 GB)23 GB free	~27 tok/s	$7.00	Deploy on RunPodfrom $0.69/hr (opens in new tab)	48 GB · 23 GB free · ~27 tok/s · $7.00 / 1M tokens
RTX 5090 (32 GB)7 GB free	~57 tok/s	$3.37	Deploy on RunPodfrom $0.69/hr (opens in new tab)	32 GB · 7 GB free · ~57 tok/s · $3.37 / 1M tokens

Compare GPUs →

Your own private copy of Qwen2.5 Coder 32B · Q4_K_M · OpenAI-compatible API

Open the llama.cpp listing on RunPod
Open on RunPodServerless (opens in new tab)
New to RunPod? Sign up and add credit first.
2
Fill in the form
Press Deploy and fill in:
Model
lmstudio-community/Qwen2.5-Coder-32B-GGUF
Quantization
Q4_K_M
Context Size
16384
After pasting the model press Esc — don’t pick a suggestion. Press Next.
Create the endpoint
Untick 24 GB, keep 48 GB, then Create Endpoint. Right away: Manage → Edit endpoint → Max workers 1 → Save (changing settings later restarts the worker).
The first start downloads the model, usually 5–15 min. After that, replies after a pause come in seconds.
Create an API key
RunPod → Settings → API Keys → Create. Pick “Restricted” with access to this endpoint only.

Use it

Copy the endpoint URL from its page on RunPod (any of its links works) and paste it with the key. Both stay in this browser.

Endpoint URL

API key

Chats and keys stay only in this browser.

https://api.runpod.ai/v2/<ENDPOINT_ID>/openai/v1

Put your endpoint ID in place of <ENDPOINT_ID>. The API key is the one from step 2.

from openai import OpenAI
client = OpenAI(base_url="https://api.runpod.ai/v2/<ENDPOINT_ID>/openai/v1", api_key="rpa_...")
r = client.chat.completions.create(
    model="any",
    messages=[{"role": "user", "content": "Hello!"}])
print(r.choices[0].message.content)

Works with any OpenAI client.

https://api.runpod.ai/v2/<ENDPOINT_ID>/openai/v1

Put your endpoint ID in place of <ENDPOINT_ID>. The API key is the one from step 2.

Free on Windows, macOS, Linux, iOS, Android and the web.

Open the web app ↗ (opens in a new tab)Download ↗ (opens in a new tab)

1. Settings → Model Provider → Add → Add Custom Provider
2. API Mode OpenAI API Compatible; paste the API key and the base URL as API Host
3. Model → Fetch, pick the model
4. New chat → pick the provider → start chatting

Free and open source for Windows, macOS and Linux.

Download Jan ↗ (opens in a new tab)

1.Settings → next to Model Providers press +
2.API format OpenAI-compatible; Base URL is the base URL above (with /v1); API key is your key; Create
3.If no model appears, click + in Models and type any
4.New chat → pick the model → start chatting

How Much VRAM Does Qwen2.5 Coder 32B Need?

Select a quantization level; the listed hardware adheres to it.

VRAM	+32K	+128K	File	Quality	On RTX 4090 (reference)
9.8 GB	17.9 GB	—	9 GB	2-bit · i-quant extreme compression — Importance-weighted 2-bit, extreme compression — significant quality loss	Fits: 32K context · ~67 tok/s
10.7 GB	18.7 GB	—	10 GB	2-bit · i-quant extra small — Importance-weighted 2-bit, extra small	Fits: 32K context · ~61 tok/s
11.1 GB	19.1 GB	—	10.4 GB	2-bit · i-quant small — Importance-weighted 2-bit, small	Fits: 32K context · ~59 tok/s
11.9 GB	19.9 GB	—	11.3 GB	2-bit · i-quant medium — Importance-weighted 2-bit, medium	Fits: 32K context · ~55 tok/s
13.5 GB	21.6 GB	—	12.8 GB	3-bit · i-quant — Importance-weighted 3-bit	Fits: 32K context · ~48 tok/s
14.3 GB	22.4 GB	—	13.7 GB	3-bit · i-quant extra small — Importance-weighted 3-bit, extra small	Fits: 32K context · ~46 tok/s
14.8 GB	22.8 GB	—	12.3 GB	2-bit · K-quant — 2-bit quantization with K-quant improvements	Fits: 32K context · ~44 tok/s
14.8 GB	22.8 GB	—	14.4 GB	3-bit · i-quant small — Importance-weighted 3-bit, small	Fits: 32K context · ~44 tok/s
15.2 GB	23.2 GB	—	14.4 GB	3-bit small — 3-bit small quantization	Fits: 32K context · ~43 tok/s
15.6 GB	23.6 GB	—	14.8 GB	3-bit · i-quant medium — Importance-weighted 3-bit, medium	Fits: 32K context · ~42 tok/s
16.8 GB	24.9 GB	—	15.9 GB	3-bit medium — 3-bit medium quantization	Fits: 16K context · ~39 tok/s
17.2 GB	25.3 GB	—	18.6 GB	4-bit · legacy — 4-bit legacy quantization	Fits: 16K context · ~38 tok/s
17.6 GB	25.7 GB	—	17.2 GB	3-bit large — 3-bit large quantization	Fits: 16K context · ~37 tok/s
18.4 GB	26.5 GB	—	17.7 GB	4-bit · i-quant compact — Importance-weighted 4-bit, compact	Fits: 16K context · ~36 tok/s
19.3 GB	27.3 GB	—	18.7 GB	4-bit · i-quant non-linear — Importance-weighted 4-bit, non-linear	Fits: 16K context · ~34 tok/s
19.3 GB	27.3 GB	—	18.8 GB	4-bit small — 4-bit small quantization	Fits: 16K context · ~34 tok/s
20.5 GB	28.6 GB	—	19.9 GB	4-bit medium — 4-bit medium quantization — most popular sweet spot	Fits: 8K context · ~32 tok/s
20.9 GB	29 GB	—	20.4 GB	4-bit large — 4-bit large quantization	Fits: 8K context · ~31 tok/s
23.4 GB	31.4 GB	—	22.6 GB	5-bit small — 5-bit small quantization	Fits: 4K context · ~28 tok/s
24.2 GB	32.2 GB	—	23.3 GB	5-bit medium — 5-bit medium quantization — good quality/size tradeoff	Does not fit: needs 27.9 GB
24.6 GB	32.6 GB	—	23.7 GB	5-bit large — 5-bit large quantization	Does not fit: needs 28.4 GB
27.9 GB	35.9 GB	—	26.9 GB	6-bit, very good — 6-bit quantization, very good quality	Does not fit: needs 31.6 GB
33.6 GB	41.7 GB	—	34.8 GB	8-bit, near-lossless — 8-bit quantization, near-lossless	Does not fit: needs 37.4 GB
66.4 GB	74.4 GB	—	~65.5 GB	16-bit · bfloat — Brain floating point 16 — preferred for training	Does not fit: needs 70.1 GB

est.calculated, no published file yet

Selected Q4_K_M

4-bit medium quantization

24.3GBat 16K context

Weights: 19.7 GB
KV cache: 4.3 GB
Overhead: 0.3 GB

Fits RTX 4090 (reference) · fits up to 8K context · ~32 tok/s

≥50 tok/s needs ≥1.57 TB/s · RTX 5090 ~57

Which Hardware Can Run Qwen2.5 Coder 32B?

Q4_K_M · 20.5 GB

Qwen2.5 Coder 32B (Q4_K_M) requires 20.5 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 27+ GB is recommended. Using the full 33K context window can add up to 8.1 GB, bringing total usage to 28.6 GB. 7 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 122 devices with unified memory can run Qwen2.5 Coder 32B, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Studio M1 Max (32 GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M

Typical hardwareQuantVRAMtok/sContext that fits

8–12 GB · RTX 4060, RTX 3060 12GBDoesn't fit — smallest quant needs 13.6 GB8 GB tier →12 GB tier →

16 GB · RTX 5060 Ti 16GBIQ2_M · 15.7 GB · ~25 tok/s · 16K15.7~2516K16 GB tier →

24 GB · RTX 4090referenceQ4_K_M · 24.3 GB · ~32 tok/s · 8Ktight24.3~328K24 GB tier →

Mac 32 GB · Mac Mini M4Q4_K_M · 24.3 GB · ~4 tok/s · 16K24.3~416KMacs →

Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 24.3 GB · ~19 tok/s · 32K24.3~1932KMacs →

How good is Qwen2.5 Coder 32B?

llmrun Score blends public benchmark boards into one number; 100 = the average open model.

Not enough data yet for an llmrun Score.

A Score needs results on enough independent boards. How the Composite Score is built

Benchmarks

MMLU
Epoch AI · Sep 2024Lab-reported
79.1%
#14/85 open
Lab-reported
GSM8K
Epoch AI · Sep 2024Lab-reported
91.1%
#4/62 open
Lab-reported
HellaSwag
Epoch AI · Sep 2024Lab-reported
83.0%
#10/47 open
Lab-reported
ARC Challenge
Epoch AI · Sep 2024Lab-reported
70.5%
#18/56 open
Lab-reported
WinoGrande
Epoch AI · Sep 2024Lab-reported
80.8%
#10/51 open
Lab-reported

All 5 Qwen2.5 Coder 32B benchmarks

Alternatives to Qwen2.5 Coder 32B

Qwen 2.5 family · 44 models →

Similar models

6tok/s on RTX 4090

Model	Score	VRAM	tok/s	Why it is similar
Nanbeige4.3 33BNanbeige	—	23.7 GB	~32	same size
MiniCPM V 4.7 35B A3BOpenBMB	—	22.1 GB	~169	same size
Llm Jp 4.1 32B A3b Thinkingllm-jp	—	20.1 GB	~188	same size
Llm Jp 4.1 33B Thinkingllm-jp	—	22.5 GB	~35	same size
Agnes 3.0 FlashAgnes-AI	—	22.1 GB	~35	same size
NVIDIA Nemotron 3.5 Lightning 30B A3B BF16NVIDIA	—	19.3 GB	~197	same size

Fine-tunes & variants(1)community builds of Qwen2.5 Coder 32B

Qwen2.5 Coder 32B Instruct

Specifications

Publisher: Alibaba
Family: Qwen 2.5
Parameters: 32.8B
Architecture: Qwen2ForCausalLM
Context Length: 32,768 tokens
Vocabulary Size: 152,064
Release Date: 2024-11-08
License: Apache 2.0

Resources

Frequently Asked Questions

How much VRAM does Qwen2.5 Coder 32B need?

Qwen2.5 Coder 32B requires 20.5 GB of VRAM at Q4_K_M, or 66.4 GB at BF16. Full 33K context adds up to 8.1 GB (28.6 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 32.8B × 4.8 bits ÷ 8 = 19.7 GB

KV Cache + Overhead ≈ 0.8 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

KV Cache + Overhead ≈ 8.9 GB (at full 33K context)

VRAM usage by quantization

Q4_K_M

20.5 GB

Q4_K_M + full context

28.6 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 4090 run Qwen2.5 Coder 32B?

Yes, at Q5_K_S (23.4 GB) or lower. Higher quantizations like Q5_K_M (24.2 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.

What's the best quantization for Qwen2.5 Coder 32B?

For Qwen2.5 Coder 32B, Q4_K_M (20.5 GB) offers the best balance of quality and VRAM usage. Q4_K_L (20.9 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 9.8 GB.

VRAM requirement by quantization

IQ2_XXS

9.8 GB~53%

Q2_K

14.8 GB~75%

Q3_K_L

17.6 GB~86%

Q4_K_M ★

20.5 GB~89%

Q4_K_L

20.9 GB~90%

BF16

66.4 GB~100%

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Qwen2.5 Coder 32B on a Mac?

Yes — MacBook Pro 14" M3 Pro (18 GB) and 108 other Macs can run Qwen2.5 Coder 32B, the smallest at IQ2_XS. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 20.5 GB of usable unified memory (RAM minus macOS overhead), from Mac Mini M5 Pro (24 GB).

Can I run Qwen2.5 Coder 32B locally?

Yes — Qwen2.5 Coder 32B can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Qwen2.5 Coder 32B?

At Q4_K_M, Qwen2.5 Coder 32B can reach ~234 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~32 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = (bandwidth GB/s ÷ model GB) × efficiency

Example: NVIDIA B200 → 8000 ÷ 20.5 × 0.65 = ~254 tok/s

Estimated speed at Q4_K_M (20.5 GB)

NVIDIA B200

~254 tok/s

NVIDIA GeForce RTX 4090

~32 tok/s

NVIDIA B300

~254 tok/s

AMD Instinct MI350X

~234 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of Qwen2.5 Coder 32B?

At Q4_K_M, the download is about 19.66 GB. The full-precision BF16 version is 65.53 GB. The smallest option (IQ2_XXS) is 9.01 GB.

Which GPUs can run Qwen2.5 Coder 32B?

7 consumer GPUs can run Qwen2.5 Coder 32B at Q4_K_M. Top options include NVIDIA GeForce RTX 5090, AMD Radeon RX 7900 XTX, NVIDIA GeForce RTX 3090. 1 GPU has plenty of headroom for comfortable inference.

Which devices can run Qwen2.5 Coder 32B?

122 devices with unified memory can run Qwen2.5 Coder 32B at Q4_K_M (20.5 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.