Best GPU for Qwen2.5 VL 72B: VRAM Requirements
Qwen2.5 VL 72B needs about 164.4 GB of GPU memory in BF16 and 43.3 GB at 4 bits. See which GPUs fit it for inference and fine-tuning.
Qwen2.5 VL 72B needs about 164.4 GB of VRAM in FP16, 83.7 GB in FP8 and 43.3 GB at 4 bits for inference with an 8,192-token context. In FP16 it fits on one NVIDIA B300 288GB; a 4-bit build fits on one NVIDIA A100 PCIe 80GB.
Qwen2.5 VL 72B is a 73.4B-parameter language model from Alibaba Cloud (Qwen team), released in January 2025, with a 32,768-token context window and published under the Qwen License. Its weights alone take 146.8 GB in BF16 and 36.7 GB at 4 bits per parameter; the tables below add the KV cache and runtime memory, then list the GPUs that hold each setup.
Recommended GPUs for Qwen2.5 VL 72B
Best for inference
1 GPU × NVIDIA H200 141GB
FP8 or INT8 weights, 83.7 GB at 8,192 tokens.
Best for fine-tuning
1 GPU × NVIDIA B300 288GB
LoRA on BF16 weights, 190.8 GB.
Budget pick
1 GPU × NVIDIA RTX PRO 6000 Blackwell 96GB
Smallest card class that holds a 4-bit build, 43.3 GB.
Qwen2.5 VL 72B VRAM requirements by precision
One request with an 8,192-token context, KV cache in BF16, plus a 10 % margin for activations and runtime buffers.
| Precision | Weights | KV cache | Total for inference |
|---|---|---|---|
| BF16 / FP16 | 146.8 GB | 2.7 GB | 164.4 GB |
| FP8 | 73.4 GB | 2.7 GB | 83.7 GB |
| INT4 | 36.7 GB | 2.7 GB | 43.3 GB |
Qwen2.5 VL 72B VRAM by task
Fine-tuning rows assume a short sequence, batch 1 and gradient checkpointing; longer sequences add activation memory.
| Task | GPU memory |
|---|---|
| Inference at the full 32,768-token context (FP8) | 92.6 GB |
| QLoRA (4-bit base) | 47.7 GB |
| LoRA (BF16 base) | 190.8 GB |
| Full fine-tuning (BF16, AdamW) | 1,291.8 GB |
GPUs that fit Qwen2.5 VL 72B
Number of cards whose combined memory holds each setup, assuming the model is split across them with tensor or pipeline parallelism. A dash means more than eight cards.
| GPU | Memory | BF16 inference (164.4 GB) | FP8 inference (83.7 GB) | INT4 inference (43.3 GB) | QLoRA (47.7 GB) | LoRA (190.8 GB) |
|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 5080 16GB | 16 GB | — | 8 GPUs | 4 GPUs | 4 GPUs | — |
| NVIDIA GeForce RTX 4090 24GB | 24 GB | 8 GPUs | 4 GPUs | 2 GPUs | 2 GPUs | 8 GPUs |
| NVIDIA L4 24GB | 24 GB | 8 GPUs | 4 GPUs | 2 GPUs | 2 GPUs | 8 GPUs |
| NVIDIA GeForce RTX 5090 32GB | 32 GB | 8 GPUs | 4 GPUs | 2 GPUs | 2 GPUs | 8 GPUs |
| NVIDIA Tesla V100 32GB | 32 GB | 8 GPUs | 4 GPUs | 2 GPUs | 2 GPUs | 8 GPUs |
| NVIDIA A100 PCIe 40GB | 40 GB | 8 GPUs | 4 GPUs | 2 GPUs | 2 GPUs | 8 GPUs |
| NVIDIA L40S 48GB | 48 GB | 4 GPUs | 2 GPUs | 1 GPU | 1 GPU | 4 GPUs |
| NVIDIA Quadro RTX 8000 48GB | 48 GB | 4 GPUs | 2 GPUs | 1 GPU | 1 GPU | 4 GPUs |
| NVIDIA RTX 6000 Ada 48GB | 48 GB | 4 GPUs | 2 GPUs | 1 GPU | 1 GPU | 4 GPUs |
| NVIDIA RTX A6000 48GB | 48 GB | 4 GPUs | 2 GPUs | 1 GPU | 1 GPU | 4 GPUs |
| NVIDIA A100 PCIe 80GB | 80 GB | 4 GPUs | 2 GPUs | 1 GPU | 1 GPU | 4 GPUs |
| NVIDIA A100 SXM 80GB | 80 GB | 4 GPUs | 2 GPUs | 1 GPU | 1 GPU | 4 GPUs |
| NVIDIA H100 PCIe 80GB | 80 GB | 4 GPUs | 2 GPUs | 1 GPU | 1 GPU | 4 GPUs |
| NVIDIA H100 SXM 80GB | 80 GB | 4 GPUs | 2 GPUs | 1 GPU | 1 GPU | 4 GPUs |
| NVIDIA RTX PRO 6000 Blackwell 96GB | 96 GB | 2 GPUs | 1 GPU | 1 GPU | 1 GPU | 2 GPUs |
| NVIDIA H200 141GB | 141 GB | 2 GPUs | 1 GPU | 1 GPU | 1 GPU | 2 GPUs |
| NVIDIA B200 180GB | 180 GB | 1 GPU | 1 GPU | 1 GPU | 1 GPU | 2 GPUs |
| NVIDIA B300 288GB | 288 GB | 1 GPU | 1 GPU | 1 GPU | 1 GPU | 1 GPU |
| NVIDIA GB300 NVL | 288 GB | 1 GPU | 1 GPU | 1 GPU | 1 GPU | 1 GPU |
What is Qwen2.5 VL 72B?
Qwen2.5 VL 72B is the largest model of the Qwen2.5 VL series, with the same image, document and long-video understanding as the smaller models and stronger results. The Qwen team also presents it as a visual agent able to operate computers and phones.
- Developer
- Alibaba Cloud (Qwen team)
- Released
- January 2025
- Parameters
- 73.4B
- Context window
- 32,768 tokens
- License
- Qwen License
The 73.4-billion-parameter total includes the vision encoder. Images and video frames use tokens of the 32,768-token window, which the model card extends with YaRN for longer inputs.
Source: Qwen2.5 VL 72B Instruct model card (as of November 17, 2026).
Frequently asked questions
- How much VRAM does Qwen2.5 VL 72B need?
- For inference with an 8,192-token context and one request, Qwen2.5 VL 72B needs about 164.4 GB in BF16, 83.7 GB in FP8, 43.3 GB in INT4, including a 10 % runtime margin. Real usage depends on the inference engine and its settings.
- Can Qwen2.5 VL 72B run on a single GPU?
- Yes. In BF16 it needs about 164.4 GB, which fits one NVIDIA B300 288GB (288 GB).
- How much GPU memory does fine-tuning Qwen2.5 VL 72B take?
- QLoRA (4-bit base): about 47.7 GB; LoRA (BF16 base): about 190.8 GB; Full fine-tuning (BF16, AdamW): about 1,291.8 GB. Longer training sequences and larger batches add activation memory on top.
- What is the best GPU for Qwen2.5 VL 72B?
- Best for inference: 1 GPU × NVIDIA H200 141GB (FP8 or INT8 weights, 83.7 GB at 8,192 tokens). Best for fine-tuning: 1 GPU × NVIDIA B300 288GB (LoRA on BF16 weights, 190.8 GB). Budget pick: 1 GPU × NVIDIA RTX PRO 6000 Blackwell 96GB (Smallest card class that holds a 4-bit build, 43.3 GB). Check live offers before booking, as availability changes.
- Can Qwen2.5 VL 72B run on an RTX 4090?
- Not on one card: even a 4-bit build needs about 43.3 GB, so it takes 2 RTX 4090 cards (48 GB combined) with the model split across them.
- Can Qwen2.5 VL 72B run on an RTX 5090?
- Not on one card: even a 4-bit build needs about 43.3 GB, so it takes 2 RTX 5090 cards (64 GB combined) with the model split across them.
Estimates use the formulas of the LLM VRAM calculator (decimal GB). Real usage depends on the inference engine, quantization format and memory fragmentation: test on the target GPU before a long rental.
Explore the live market
Every offer on the live market comes from a reviewed supplier and is sold by EverXCloud Inc.