Qwen · 32B to 80B
Qwen2.5 VL 72B Instruct
73.41 billion parameters across 80 layers. At four-bit precision the weights alone come to 41.28 GB, before any conversation is loaded. It uses grouped-query attention with 8 KV heads for 64 query heads, which cuts the cache to 13% of what full attention would need.
About Qwen2.5 VL 72B Instruct
Qwen2.5 VL 72B Instruct is a dense model with 73.41B parameters, built from 80 layers with a hidden size of 8192. Its repository was published on 2025-01-27. The license listed on Hugging Face is other. It uses grouped-query attention with 8 KV heads for 64 query heads, which cuts the cache to 13% of what full attention would need.
Memory needed, by quantization
At an 8k context window. Quality is our rough ranking of how much the compression costs you: anything at or above Q5 is hard to tell apart from the original in normal use.
| Quantization | File size | KV cache | Total VRAM | Quality | What it costs you |
|---|---|---|---|---|---|
| Q8_0 | 72.64 GB | 2.5 GB | 74.45 GB | 99% | Lossless in practice. Use it when the memory is there. |
| Q6_K | 56.06 GB | 2.5 GB | 58.15 GB | 98% | Very close to Q8 for two thirds of the size. |
| Q5_K_M | 48.63 GB | 2.5 GB | 50.84 GB | 96% | The quality-first choice when Q6 will not fit. |
| Q4_K_M | 41.28 GB | 2.5 GB | 43.62 GB | 93% | The default. Best size-to-quality ratio for local use. |
| IQ4_XS | 36.32 GB | 2.5 GB | 38.75 GB | 91% | Importance-matrix 4-bit. Q4_K_S quality, smaller file. |
| Q3_K_M | 33.42 GB | 2.5 GB | 35.89 GB | 86% | Degradation starts to show. A way to fit one size up. |
| IQ3_XXS | 26.15 GB | 2.5 GB | 28.75 GB | 79% | Aggressive. Only worth it on very large models. |
What a longer conversation costs
Same model at Q4_K_M, only the context window changes. The cache grows in a straight line with every token in the window.
| Context | KV cache | Total VRAM | Fits in 8 GB | Fits in 12 GB | Fits in 24 GB |
|---|---|---|---|---|---|
| 4k | 1.25 GB | 42.37 GB | no | no | no |
| 8k | 2.5 GB | 43.62 GB | no | no | no |
| 16k | 5 GB | 46.12 GB | no | no | no |
| 32k | 10 GB | 51.12 GB | no | no | no |
| 64k | 20 GB | 61.12 GB | no | no | no |
The "fits" columns allow for the roughly 0.8 GB Windows keeps for the desktop.
Which hardware runs Qwen2.5 VL 72B Instruct
15 of 118 consumer devices run it at a quantization worth using, at an 8k context window. Another 7 can load it only by compressing the weights far enough to damage the model, marked with a warning below.
| Device | Memory | Verdict | Quantization | Used | Speed |
|---|---|---|---|---|---|
| RTX PRO 6000 Blackwell NVIDIA | 96 GB | Runs well | Q8_0 | 74.45 GB | 17.8 t/s |
| RTX 6000 Ada Generation NVIDIA | 48 GB | Runs well | Q4_K_M | 43.62 GB | 16.4 t/s |
| RTX A6000 NVIDIA | 48 GB | Runs well | Q4_K_M | 43.62 GB | 13.1 t/s |
| Apple M3 Ultra 96GB Apple | 96 GB | Runs well | Q5_K_M | 51.64 GB | 11.1 t/s |
| Apple M3 Ultra 256GB Apple | 256 GB | Runs well | Q5_K_M | 51.64 GB | 11.1 t/s |
| Apple M3 Ultra 512GB Apple | 512 GB | Runs well | Q5_K_M | 51.64 GB | 11.1 t/s |
| Apple M1 Ultra 128GB Apple | 128 GB | Runs well | Q5_K_M | 51.64 GB | 10.8 t/s |
| Apple M2 Ultra 128GB Apple | 128 GB | Runs well | Q5_K_M | 51.64 GB | 10.8 t/s |
| Apple M2 Ultra 192GB Apple | 192 GB | Runs well | Q5_K_M | 51.64 GB | 10.8 t/s |
| Radeon PRO W7900 AMD | 48 GB | Runs well | Q4_K_M | 43.62 GB | 12.5 t/s |
| GeForce RTX 5090 NVIDIA | 32 GB | Runs great | IQ3_XXS ! | 28.75 GB | 46.7 t/s |
| Apple M4 Max 64GB Apple | 64 GB | Runs well | Q3_K_M | 36.43 GB | 10.5 t/s |
| Apple M4 Max 128GB Apple | 128 GB | Runs well | Q3_K_M | 36.43 GB | 10.5 t/s |
| Apple M3 Max 128GB Apple | 128 GB | Fits, but slow | Q8_0 | 75.65 GB | 3.7 t/s |
| NVIDIA DGX Spark 128GB NVIDIA | 128 GB | Fits, but slow | Q8_0 | 75.68 GB | 2.7 t/s |
| Ryzen AI Max+ 395 128GB AMD | 128 GB | Fits, but slow | Q8_0 | 75.68 GB | 2 t/s |
| GeForce RTX 4090 NVIDIA | 24 GB | Runs great | IQ2_XXS ! | 20.35 GB | 37.4 t/s |
| GeForce RTX 3090 Ti NVIDIA | 24 GB | Runs great | IQ2_XXS ! | 20.35 GB | 37.4 t/s |
| GeForce RTX 3090 NVIDIA | 24 GB | Runs great | IQ2_XXS ! | 20.35 GB | 34.8 t/s |
| GeForce RTX 5090 Laptop NVIDIA | 24 GB | Runs great | IQ2_XXS ! | 20.35 GB | 33.3 t/s |
| Radeon RX 7900 XTX AMD | 24 GB | Runs great | IQ2_XXS ! | 20.35 GB | 30.2 t/s |
| RTX A5000 NVIDIA | 24 GB | Runs great | IQ2_XXS ! | 20.35 GB | 28.5 t/s |
| Apple M4 Max 36GB Apple | 36 GB | Just fits | n/a ! | 29.16 GB | 13.2 t/s |
| Apple M1 Ultra 64GB Apple | 64 GB | Just fits | n/a | 51.64 GB | 10.8 t/s |
| Apple M2 Ultra 64GB Apple | 64 GB | Just fits | n/a | 51.64 GB | 10.8 t/s |
Show 35 more devices
| Apple M4 Max 48GB Apple | 48 GB | Just fits | n/a | 36.43 GB | 10.5 t/s |
| Apple M1 Max 32GB Apple | 32 GB | Just fits | n/a ! | 29.16 GB | 9.6 t/s |
| Apple M2 Max 32GB Apple | 32 GB | Just fits | n/a ! | 29.16 GB | 9.6 t/s |
| Apple M3 Max 36GB Apple | 36 GB | Just fits | n/a ! | 29.16 GB | 9.6 t/s |
| Apple M4 Pro 24GB Apple | 24 GB | Just fits | n/a ! | 20.62 GB | 9.4 t/s |
| Apple M3 Max 48GB Apple | 48 GB | Just fits | n/a | 39.33 GB | 7.1 t/s |
| Apple M1 Max 64GB Apple | 64 GB | Just fits | n/a | 51.64 GB | 5.4 t/s |
| Apple M2 Max 64GB Apple | 64 GB | Just fits | n/a | 51.64 GB | 5.4 t/s |
| Apple M3 Max 64GB Apple | 64 GB | Just fits | n/a | 51.64 GB | 5.4 t/s |
| Apple M5 24GB Apple | 24 GB | Just fits | n/a ! | 20.62 GB | 5.3 t/s |
| Jetson AGX Orin 32GB NVIDIA | 32 GB | Just fits | n/a ! | 29.19 GB | 5.3 t/s |
| Ryzen AI Max+ 395 32GB AMD | 32 GB | Just fits | n/a ! | 29.19 GB | 5.2 t/s |
| Apple M4 Pro 48GB Apple | 48 GB | Just fits | n/a | 39.33 GB | 4.9 t/s |
| Apple M1 Pro 32GB Apple | 32 GB | Just fits | n/a ! | 29.16 GB | 4.8 t/s |
| Apple M2 Pro 32GB Apple | 32 GB | Just fits | n/a ! | 29.16 GB | 4.8 t/s |
| Apple M4 24GB Apple | 24 GB | Just fits | n/a ! | 20.62 GB | 4.1 t/s |
| Apple M2 Max 96GB Apple | 96 GB | Just fits | n/a | 75.65 GB | 3.7 t/s |
| Apple M3 Max 96GB Apple | 96 GB | Just fits | n/a | 75.65 GB | 3.7 t/s |
| Apple M4 Pro 64GB Apple | 64 GB | Just fits | n/a | 51.64 GB | 3.7 t/s |
| Apple M5 32GB Apple | 32 GB | Just fits | n/a ! | 29.16 GB | 3.7 t/s |
| Ryzen AI 9 HX 370 24GB AMD | 24 GB | Just fits | n/a ! | 20.65 GB | 3.7 t/s |
| Apple M3 Pro 36GB Apple | 36 GB | Just fits | n/a ! | 29.16 GB | 3.6 t/s |
| Apple M2 24GB Apple | 24 GB | Just fits | n/a ! | 20.62 GB | 3.4 t/s |
| Apple M3 24GB Apple | 24 GB | Just fits | n/a ! | 20.62 GB | 3.4 t/s |
| Jetson AGX Orin 64GB NVIDIA | 64 GB | Just fits | n/a | 51.67 GB | 3 t/s |
| Apple M4 32GB Apple | 32 GB | Just fits | n/a ! | 29.16 GB | 2.9 t/s |
| Ryzen AI Max+ 395 64GB AMD | 64 GB | Just fits | n/a | 51.67 GB | 2.9 t/s |
| Intel Core Ultra 9 288V 32GB Intel | 32 GB | Just fits | n/a ! | 29.19 GB | 2.7 t/s |
| Ryzen AI 9 HX 370 32GB AMD | 32 GB | Just fits | n/a ! | 29.19 GB | 2.6 t/s |
| Ryzen AI Max+ 395 96GB AMD | 96 GB | Just fits | n/a | 75.68 GB | 2 t/s |
| Radeon RX 7900 XT AMD | 20 GB | Partial offload 34/80 | n/a | 43.62 GB | 1.2 t/s |
| GeForce RTX 5080 NVIDIA | 16 GB | Partial offload 27/80 | n/a | 43.62 GB | 1 t/s |
| GeForce RTX 5070 Ti NVIDIA | 16 GB | Partial offload 27/80 | n/a | 43.62 GB | 1 t/s |
| GeForce RTX 5060 Ti 16GB NVIDIA | 16 GB | Partial offload 27/80 | n/a | 43.62 GB | 1 t/s |
| GeForce RTX 4080 SUPER NVIDIA | 16 GB | Partial offload 27/80 | n/a | 43.62 GB | 1 t/s |
How it compares
Other models in the same family or of a similar size, on a 12 GB card and a 24 GB card, each at the best quantization that fits.
| Model | Weights at 4 bits | GeForce RTX 3060 12GB | GeForce RTX 4090 |
|---|---|---|---|
| Qwen2.5 VL 72B Instruct | 41.3 GB | does not fit well | does not fit well |
| Qwen2.5 72B Instruct | 40.9 GB | does not fit well | does not fit well |
| Qwen3 Coder Next | 44.8 GB | 33.2 t/s at Q4_K_M | 56.4 t/s at Q4_K_M |
| Qwen3 Next 80B A3B Instruct | 45.7 GB | 32.7 t/s at Q4_K_M | 54.6 t/s at Q4_K_M |
| Llama 3.3 70B Instruct | 39.7 GB | does not fit well | does not fit well |
What to buy to run Qwen2.5 VL 72B Instruct
The cheapest hardware that runs it at a quantization worth using and a speed you would not resent, at an 8k context window.
Cheapest that works
RTX 6000 Ada Generation
$6,800
Q4_K_M · uses 43.62 GB of its 48 GB · about 16.4 tokens/s
Best value
RTX PRO 6000 Blackwell
$8,500
Q8_0 · about 17.8 tokens/s · 1792 GB/s
If it just has to run
Ryzen AI Max+ 395 64GB
$1,699 whole machine
Q3_K_M · about 4.1 tokens/s. Cheaper because it runs the model at a lower precision.
| Device | Price | Memory | Runs it at | Speed | |
|---|---|---|---|---|---|
| RTX 6000 Ada Generation NVIDIA | $6,800 | 48 GB | Q4_K_M | 16.4 t/s | Check price |
| RTX PRO 6000 Blackwell NVIDIA | $8,500 | 96 GB | Q8_0 | 17.8 t/s | Check price |
Indicative prices reviewed 2026-09-01; used prices are marketplace typical. Options that cost more than a cheaper one with no more memory and no more speed are hidden. Price links are Amazon affiliate links. Change the standard, the context window or the budget in the buying tool.
Where it sits in the catalog
Source: Qwen/Qwen2.5-VL-72B-Instruct on Hugging Face. Downloaded 223k times in the last month. Published 2025-01-27. Architecture figures are read from the repository's own configuration file, so they move when the model does.
Qwen2.5 72B Instruct
72.71B · 32k context
Qwen3 Coder Next
79.67B (3.19B active) · 256k context
Qwen3 Next 80B A3B Instruct
81.32B (3.19B active) · 256k context
Qwen3.5 35B A3B
35.95B (2.9B active) · 256k context
Qwen3 32B
32.76B · 40k context
Qwen2.5 32B Instruct
32.76B · 32k context
Qwen2.5 Coder 32B Instruct
32.76B · 32k context
Qwen3 30B A3B
30.53B (3.34B active) · 40k context
Questions people ask about Qwen2.5 VL 72B Instruct
How much VRAM does Qwen2.5 VL 72B Instruct need?
At Q4_K_M with an 8k context window, Qwen2.5 VL 72B Instruct needs about 43.62 GB: 40.58 GB of weights, 2.5 GB of KV cache and 0.54 GB of overhead. At the highest-quality file it needs about 74.45 GB.
What is the cheapest GPU that runs Qwen2.5 VL 72B Instruct?
The smallest consumer device in our catalog that fits it is the Apple M4 Max 64GB with 64 GB, running it at Q3_K_M at roughly 10.5 tokens per second.
How much does a longer context window add for Qwen2.5 VL 72B Instruct?
On the memory side, a 32k window adds 10.0 GB of KV cache on top of the weights, and the cache grows in a straight line with every token. The longest window the model supports natively is 125k tokens.