How much memory do I need to run Qwen2.5-72B-Instruct (Q4_K_M)?

The file size is 47.4 GB (measured). With a 8,192-token context, the estimated memory including the KV cache is 50.9 GB. The formula is published at https://localllm.az-arte.com/en/guide/memory-calculation.

Does Qwen2.5-72B-Instruct run in 8 GB, 12 GB, 16 GB or 24 GB of VRAM?

8 GB: no quantization fits; 12 GB: no quantization fits; 16 GB: no quantization fits; 24 GB: no quantization fits (estimated at 8,192 tokens with an f16 KV cache, files from the most-downloaded source).

What is the command to run Qwen2.5-72B-Instruct with Ollama?

Run "ollama run hf.co/bartowski/Qwen2.5-72B-Instruct-GGUF:Q4_K_M" to use the GGUF on Hugging Face directly. To use another quantization, replace the name at the end.

Qwen

Qwen2.5-72B-Instruct

Parameters: 72.7B
Max context length: 32,768
Architecture: qwen2
License: Conditional

Base model: Qwen/Qwen2.5-72B-Instruct
License: qwen

Memory needed to run Q4_K_M at 8K tokens

50.9 GB Estimated

File size 47.4 GB Measured

VRAM it fits in (rough guide)

8 GB does not fit
12 GB does not fit
16 GB does not fit
24 GB does not fit
32 GB does not fit
48 GB fits

See devices and models for 48 GB

The smallest, IQ1_M, needs an estimated 27.3 GB at 8K.

Macs cap how much memory the GPU can use, so leave headroom in unified memory.

Data as of Nov 11, 2026

About Qwen2.5-72B-Instruct

Qwen2.5-72B-Instruct is a GGUF model based on Qwen/Qwen2.5-72B-Instruct on Hugging Face. GGUF files are published by bartowski/Qwen2.5-72B-Instruct-GGUF. It has about 72.7B parameters, the qwen2 architecture, a maximum context length of 32,768 tokens (values written in the GGUF).

There are 18 quantizations, with file sizes from 23.7 GB (IQ1_M) to 77.3 GB (Q8_0) (actual sizes on Hugging Face).

Running Q4_K_M (47.4 GB) with an 8,192-token context needs an estimated 50.9 GB of memory, and an estimated 59.0 GB at 32,768 tokens. For 8,192 tokens, a GPU with 48 GB or more of VRAM, or a Mac with 48 GB or more of unified memory, is a rough guide (macOS caps how much memory the GPU can use, so leave some headroom). The smallest, IQ1_M (23.7 GB), needs an estimated 27.3 GB at 8,192 tokens.

The original model is licensed under Qwen License Agreement, which allows commercial use with conditions such as scale or purpose. Check the original license for the terms of use.

File size and estimated memory per quantization

Up to 8 GB
Up to 16 GB
Up to 24 GB
Up to 48 GB
Up to 128 GB
More

File sizes are the actual values on Hugging Face. Memory is estimated for each context length (KV cache f16). The formula is in How memory is calculated.

All memory figures are Estimated (KV cache f16). It should run on a setup with at least this much VRAM (or Mac unified memory). Click a value to see the devices and models that fit that amount (Find by GPU / Mac).

About the "quality guide" (how far a quantization drifts from the original)

For each quantization type, how far the output drifts from the original model, grouped into six levels from values llama.cpp measured on Llama-3-8B.

The value is mean KLD (the difference between the original and quantized model in next-token probabilities); smaller means closer to the original. Quantizations made with an imatrix (importance data) come out slightly smaller. Levels use the value without an imatrix (or with one, for types that only have that).

These values were not measured on this model. Larger models and MoE models tend to drift less with the same type. Distributor-specific mixed types (such as UD-Q4_K_XL) cannot be matched by type name alone, so they show "-".

Quantization	Quality guide	KLD (no imatrix)	KLD (imatrix)	Size (Llama-3-8B)
Q8_0	Nearly identical	0.0014	-	7.96 GiB
Q6_K	Very small	0.0055	-	6.14 GiB
Q5_K_M	Very small	0.0108	-	5.33 GiB
Q5_K_S	Small	0.0166	-	5.21 GiB
Q5_1	Small	0.0180	-	5.65 GiB
Q5_0	Small	0.0222	-	5.21 GiB
Q4_K_M	Small	0.0313	0.0282	4.58 GiB
Q4_K_S	Medium	0.0431	0.0320	4.37 GiB
IQ4_NL	Small	-	0.0357	4.35 GiB
IQ4_XS	Small	-	0.0363	4.14 GiB
Q4_1	Medium	0.0717	-	4.78 GiB
Q4_0	Medium	0.0719	-	4.34 GiB
Q3_K_L	Medium	0.0872	0.0731	4.03 GiB
Q3_K_M	Medium	0.1019	0.0844	3.74 GiB
IQ3_M	Medium	-	0.1025	3.53 GiB
IQ3_S	Medium	-	0.1113	3.42 GiB
IQ3_XS	Large	-	0.1387	3.28 GiB
IQ3_XXS	Large	-	0.1836	3.05 GiB
Q3_K_S	Large	0.2282	0.1998	3.41 GiB
IQ2_M	Large	-	0.3260	2.74 GiB
Q2_K	Very large	0.4451	0.3322	2.96 GiB
Q2_K_S	Very large	-	0.4034	2.96 GiB
IQ2_S	Very large	-	0.4392	2.56 GiB
IQ2_XS	Very large	-	0.5463	2.43 GiB
IQ2_XXS	Very large	-	0.8120	2.24 GiB
IQ1_M	Very large	-	1.3931	2.01 GiB
IQ1_S	Very large	-	2.2546	1.88 GiB

Source: llama.cpp (MIT), measured on Llama-3-8B against FP16, Wikitext-2, revision f364eb6f. Original table · Reading and choosing quantizations

GGUF source

bartowski/Qwen2.5-72B-Instruct-GGUF

Converted and published by a third party About this label

Last checked: Nov 11, 2026

Architecture: qwen2; Supported before llama.cpp b4409
Parameters: 72.7B
Max context length: 32,768
License: qwen Conditional

File size and memory per quantization for Qwen2.5-72B-Instruct (bartowski/Qwen2.5-72B-Instruct-GGUF)
Quantization	File size Measured	Quality guide	4,096	8,192	16,384	32,768 max
IQ1_M	23.7 GB	Very large	25.9 GB	27.3 GB	30.0 GB	35.3 GB
IQ2_XXS	25.5 GB	Very large	27.7 GB	29.0 GB	31.7 GB	37.1 GB
IQ2_XS	27.1 GB	Very large	29.2 GB	30.6 GB	33.3 GB	38.6 GB
IQ2_M	29.3 GB	Large	31.5 GB	32.9 GB	35.6 GB	40.9 GB
Q2_K	29.8 GB	Very large	32.0 GB	33.3 GB	36.0 GB	41.4 GB
Q2_K_L	31.0 GB	-	33.2 GB	34.6 GB	37.2 GB	42.6 GB
IQ3_XXS	31.8 GB	Large	34.0 GB	35.4 GB	38.1 GB	43.4 GB
Q3_K_S	34.5 GB	Large	36.7 GB	38.0 GB	40.7 GB	46.1 GB
IQ3_M	35.5 GB	Medium	37.7 GB	39.0 GB	41.7 GB	47.1 GB
Q3_K_M	37.7 GB	Medium	39.9 GB	41.2 GB	43.9 GB	49.3 GB
Q3_K_L	39.5 GB	Medium	41.7 GB	43.0 GB	45.7 GB	51.1 GB
IQ4_XS	39.7 GB	Small	41.9 GB	43.2 GB	45.9 GB	51.3 GB
Q3_K_XL	40.6 GB	-	42.8 GB	44.1 GB	46.8 GB	52.2 GB
Q4_0	41.4 GB	Medium	43.6 GB	44.9 GB	47.6 GB	53.0 GB
Q4_K_M	47.4 GB	Small	49.6 GB	50.9 GB	53.6 GB	59.0 GB
Q5_K_M (2 parts)	54.4 GB	Very small	56.6 GB	58.0 GB	60.7 GB	66.0 GB
Q6_K (2 parts)	64.3 GB	Very small	66.5 GB	67.9 GB	70.6 GB	75.9 GB
Q8_0 (2 parts)	77.3 GB	Nearly identical	79.5 GB	80.8 GB	83.5 GB	88.8 GB

Ollama  ollama run hf.co/bartowski/Qwen2.5-72B-Instruct-GGUF:Q4_K_M

llama.cpp  llama-server -hf bartowski/Qwen2.5-72B-Instruct-GGUF:Q4_K_M

Generation speed on common GPUs and Macs (estimated)

A rough guide to how fast Q4_K_M (47.4 GB) generates text. Each token reads the model weights once, so the upper limit is memory bandwidth ÷ file size. The guide is 60–80% of that, where most published measurements fall.

Device	Memory bandwidth	Guide (tokens/s) Estimated	Upper limit
Apple M5 Pro 64 GB	307 GB/s	3.9–5.2	6.5
Apple M2 Max 64 GB	400 GB/s	5.1–6.7	8.4
Apple M5 Ultra 96 GB	1,200 GB/s	15–20	25

Estimates only. Prompt processing speed (limited by compute) is not included. Longer contexts and offloading part of the model to the CPU make it slower. Macs are shown only when installed memory is at least 4/3 of what is needed. Devices without published bandwidth (some GeForce cards, etc.) are not shown.

Third-party evaluations

Values from other organizations for the original model before quantization. Not this site's own rating. Values are for the original model before quantization. GGUF quantization (especially small quantizations) can lower performance.

Human preference (LMArena)

1,303 1,299–1,307

qwen2.5-72b-instruct · 39,406 votes · Values as of Nov 9, 2026

Source: LMArena (CC BY 4.0). Original leaderboard · See in ranking

Hard science (GPQA)

49.1% 43.9%–54.4%

qwen2.5-72b-instruct · Values as of Jan 27, 2025

Source: GPQA Diamond (Epoch AI) (CC BY 4.0). Original leaderboard · See in ranking

License and commercial use

Qwen License Agreement Conditional

Allows commercial use. However, products or services with more than 100 million monthly active users need a separate license from Alibaba Cloud.

Based on the license on the original model's Hugging Face page. Original license text (checked Oct 18, 2026)

The summary states facts this site read from the original text and is not legal advice. Always check the original license and its latest terms before use. All licenses

How to run

Example commands using Q4_K_M. To use another quantization, replace the name with one from the table above. How to run local LLMs

Ollama

Run it by pointing directly at the GGUF on Hugging Face.

ollama run hf.co/bartowski/Qwen2.5-72B-Instruct-GGUF:Q4_K_M

llama.cpp

Pass the repository and quantization name to llama-server (or llama-cli) with -hf.

llama-server -hf bartowski/Qwen2.5-72B-Instruct-GGUF:Q4_K_M

LM Studio

Paste bartowski/Qwen2.5-72B-Instruct-GGUF into the search box, download it, and choose the quantization from the file list.

Frequently asked questions

How much memory do I need to run Qwen2.5-72B-Instruct (Q4_K_M)?: The file size is 47.4 GB (measured). With a 8,192-token context, the estimated memory including the KV cache is 50.9 GB. The formula is published at https://localllm.az-arte.com/en/guide/memory-calculation.
Can Qwen2.5-72B-Instruct be used commercially?: The original model's license is Qwen License Agreement (Commercial use with conditions).Allows commercial use. However, products or services with more than 100 million monthly active users need a separate license from Alibaba Cloud. Check the original license text before deciding.
Does Qwen2.5-72B-Instruct run in 8 GB, 12 GB, 16 GB or 24 GB of VRAM?: 8 GB: no quantization fits; 12 GB: no quantization fits; 16 GB: no quantization fits; 24 GB: no quantization fits (estimated at 8,192 tokens with an f16 KV cache, files from the most-downloaded source).
What is the command to run Qwen2.5-72B-Instruct with Ollama?: Run "ollama run hf.co/bartowski/Qwen2.5-72B-Instruct-GGUF:Q4_K_M" to use the GGUF on Hugging Face directly. To use another quantization, replace the name at the end.