Best GPU for Popular AI Models: VRAM by Model
How much GPU memory Llama, Qwen, DeepSeek, Kimi, GLM, Mistral, Gemma, Phi, gpt-oss, FLUX.1, Stable Diffusion, Wan and Whisper need, and which GPUs hold them.
Each page computes the GPU memory of one open-weight model from its published architecture: weights in BF16, FP8 and 4-bit, KV cache at an 8,192-token context, LoRA, QLoRA and full fine-tuning. It then lists the GPUs that hold each setup on one to eight cards.
Llama (Meta)
- Best GPU for Llama 3.2 1B
1.2B parameters · Meta
BF16 3 GB · 4-bit 1 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Llama 3.2 3B
3.2B parameters · Meta
BF16 8.1 GB · 4-bit 2.8 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Llama 3.1 8B
8B parameters · Meta
BF16 18.8 GB · 4-bit 5.6 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Llama Guard 3 8B
8B parameters · Meta
BF16 18.8 GB · 4-bit 5.6 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Llama 3.2 11B Vision
10.7B parameters · Meta
BF16 25 GB · 4-bit 7.4 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Llama 3.3 70B
70.6B parameters · Meta
BF16 158.3 GB · 4-bit 41.8 GB
Inference: 1 GPU × NVIDIA H200 141GB
- Best GPU for Llama 4 Scout
109B parameters · Meta
BF16 241.6 GB · 4-bit 61.7 GB
Inference: 1 GPU × NVIDIA H200 141GB
- Best GPU for Llama 4 Maverick
400B parameters · Meta
BF16 881.8 GB · 4-bit 221.8 GB
Inference: 2 GPUs × NVIDIA B300 288GB
- Best GPU for Llama 3.1 405B
405B parameters · Meta
BF16 895.7 GB · 4-bit 227.4 GB
Inference: 2 GPUs × NVIDIA B300 288GB
Qwen (Alibaba Cloud)
- Best GPU for Qwen3 4B
4B parameters · Alibaba Cloud (Qwen team)
BF16 10.2 GB · 4-bit 3.5 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Qwen2.5 7B
7.6B parameters · Alibaba Cloud (Qwen team)
BF16 17.3 GB · 4-bit 4.7 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Qwen3 8B
8.2B parameters · Alibaba Cloud (Qwen team)
BF16 19.4 GB · 4-bit 5.8 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Qwen2.5 VL 7B
8.3B parameters · Alibaba Cloud (Qwen team)
BF16 18.8 GB · 4-bit 5.1 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Qwen2.5 14B
14.7B parameters · Alibaba Cloud (Qwen team)
BF16 34.1 GB · 4-bit 9.9 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Qwen3 14B
14.8B parameters · Alibaba Cloud (Qwen team)
BF16 34 GB · 4-bit 9.6 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Qwen3 30B A3B
30.5B parameters · Alibaba Cloud (Qwen team)
BF16 68 GB · 4-bit 17.7 GB
Inference: 1 GPU × NVIDIA A100 PCIe 40GB
- Best GPU for Qwen3 Coder 30B A3B
30.5B parameters · Alibaba Cloud (Qwen team)
BF16 68 GB · 4-bit 17.7 GB
Inference: 1 GPU × NVIDIA A100 PCIe 40GB
- Best GPU for Qwen2.5 Coder 32B
32.5B parameters · Alibaba Cloud (Qwen team)
BF16 73.9 GB · 4-bit 20.2 GB
Inference: 1 GPU × NVIDIA L40S 48GB
- Best GPU for Qwen3 32B
32.8B parameters · Alibaba Cloud (Qwen team)
BF16 74.5 GB · 4-bit 20.4 GB
Inference: 1 GPU × NVIDIA L40S 48GB
- Best GPU for Qwen2.5 72B
72.7B parameters · Alibaba Cloud (Qwen team)
BF16 162.9 GB · 4-bit 42.9 GB
Inference: 1 GPU × NVIDIA H200 141GB
- Best GPU for Qwen2.5 VL 72B
73.4B parameters · Alibaba Cloud (Qwen team)
BF16 164.4 GB · 4-bit 43.3 GB
Inference: 1 GPU × NVIDIA H200 141GB
- Best GPU for Qwen3 235B A22B
235B parameters · Alibaba Cloud (Qwen team)
BF16 518.7 GB · 4-bit 131 GB
Inference: 2 GPUs × NVIDIA B200 180GB
DeepSeek
- Best GPU for DeepSeek R1 Distill Qwen 7B
7.6B parameters · DeepSeek
BF16 17.3 GB · 4-bit 4.7 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for DeepSeek R1 Distill Llama 8B
8B parameters · DeepSeek
BF16 18.8 GB · 4-bit 5.6 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for DeepSeek R1 Distill Qwen 14B
14.8B parameters · DeepSeek
BF16 34.3 GB · 4-bit 9.9 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for DeepSeek R1 Distill Qwen 32B
32.8B parameters · DeepSeek
BF16 74.5 GB · 4-bit 20.4 GB
Inference: 1 GPU × NVIDIA L40S 48GB
- Best GPU for DeepSeek R1 Distill Llama 70B
70.6B parameters · DeepSeek
BF16 158.3 GB · 4-bit 41.8 GB
Inference: 1 GPU × NVIDIA H200 141GB
- Best GPU for DeepSeek R1
671B parameters · DeepSeek
BF16 1,476.8 GB · 4-bit 369.7 GB
Inference: 4 GPUs × NVIDIA B300 288GB
- Best GPU for DeepSeek V3
671B parameters · DeepSeek
BF16 1,476.8 GB · 4-bit 369.7 GB
Inference: 4 GPUs × NVIDIA B300 288GB
Kimi (Moonshot AI)
- Best GPU for Kimi K2
1,026B parameters · Moonshot AI
BF16 2,257.8 GB · 4-bit 564.9 GB
Inference: 8 GPUs × NVIDIA B200 180GB
GLM (Z.ai)
- Best GPU for GLM-4.5 Air
106B parameters · Z.ai
BF16 234.9 GB · 4-bit 60 GB
Inference: 1 GPU × NVIDIA H200 141GB
- Best GPU for GLM-4.5
355B parameters · Z.ai
BF16 784.4 GB · 4-bit 198.6 GB
Inference: 2 GPUs × NVIDIA B300 288GB
Mistral AI
- Best GPU for Mistral 7B v0.3
7.3B parameters · Mistral AI
BF16 17.1 GB · 4-bit 5.2 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Mistral NeMo 12B
12.2B parameters · Mistral AI and NVIDIA
BF16 28.3 GB · 4-bit 8.2 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Codestral 22B
22.2B parameters · Mistral AI
BF16 50.9 GB · 4-bit 14.3 GB
Inference: 1 GPU × NVIDIA A100 PCIe 40GB
- Best GPU for Mistral Small 3 (24B)
23.6B parameters · Mistral AI
BF16 53.4 GB · 4-bit 14.5 GB
Inference: 1 GPU × NVIDIA A100 PCIe 40GB
- Best GPU for Mistral Small 3.1 (24B)
24B parameters · Mistral AI
BF16 54.3 GB · 4-bit 14.7 GB
Inference: 1 GPU × NVIDIA A100 PCIe 40GB
- Best GPU for Mixtral 8x7B
46.7B parameters · Mistral AI
BF16 103.9 GB · 4-bit 26.9 GB
Inference: 1 GPU × NVIDIA A100 PCIe 80GB
- Best GPU for Mistral Large 2 (123B)
123B parameters · Mistral AI
BF16 273.8 GB · 4-bit 70.9 GB
Inference: 1 GPU × NVIDIA B200 180GB
Gemma (Google)
- Best GPU for Gemma 3 4B
4.3B parameters · Google DeepMind
BF16 10.7 GB · 4-bit 3.6 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Gemma 2 9B
9.2B parameters · Google DeepMind
BF16 23.4 GB · 4-bit 8.2 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Gemma 3 12B
12.2B parameters · Google DeepMind
BF16 30.4 GB · 4-bit 10.3 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Gemma 3 27B
27.4B parameters · Google DeepMind
BF16 64.9 GB · 4-bit 19.6 GB
Inference: 1 GPU × NVIDIA A100 PCIe 40GB
Phi (Microsoft)
- Best GPU for Phi-4 mini (3.8B)
3.8B parameters · Microsoft
BF16 9.6 GB · 4-bit 3.3 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Phi-4 (14B)
14.7B parameters · Microsoft
BF16 34.2 GB · 4-bit 9.9 GB
Inference: 1 GPU × NVIDIA L4 24GB
gpt-oss (OpenAI)
- Best GPU for gpt-oss-20b
20.9B parameters · OpenAI
BF16 46.4 GB · 4-bit 11.9 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for gpt-oss-120b
116.8B parameters · OpenAI
BF16 257.6 GB · 4-bit 64.9 GB
Inference: 1 GPU × NVIDIA A100 PCIe 80GB
Image generation
- Best GPU for Stable Diffusion XL 1.0
3.5B parameters · Stability AI
BF16 7.7 GB · 4-bit 1.9 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Stable Diffusion 3.5 Large
13.8B parameters · Stability AI
BF16 30.4 GB · 4-bit 7.6 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for FLUX.1 [dev]
16.9B parameters · Black Forest Labs
BF16 37.2 GB · 4-bit 9.3 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for FLUX.1 [schnell]
16.9B parameters · Black Forest Labs
BF16 37.2 GB · 4-bit 9.3 GB
Inference: 1 GPU × NVIDIA L4 24GB
Video generation
- Best GPU for Wan 2.1 1.3B
7.2B parameters · Alibaba (Wan team)
BF16 15.9 GB · 4-bit 4 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for CogVideoX 5B
10.5B parameters · Zhipu AI and Tsinghua University
BF16 23.1 GB · 4-bit 5.8 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Wan 2.1 14B
20.1B parameters · Alibaba (Wan team)
BF16 44.2 GB · 4-bit 11.1 GB
Inference: 1 GPU × NVIDIA A100 PCIe 40GB
- Best GPU for HunyuanVideo
20.7B parameters · Tencent
BF16 45.5 GB · 4-bit 11.4 GB
Inference: 1 GPU × NVIDIA A100 PCIe 40GB
Speech recognition and embeddings
- Best GPU for BGE-M3
0.6B parameters · BAAI
BF16 1.2 GB · 4-bit 0.3 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Whisper large-v3 turbo
0.8B parameters · OpenAI
BF16 1.8 GB · 4-bit 0.4 GB
Inference: 1 GPU × NVIDIA L4 24GB
- Best GPU for Whisper large-v3
1.6B parameters · OpenAI
BF16 3.4 GB · 4-bit 0.9 GB
Inference: 1 GPU × NVIDIA L4 24GB
Explore the live market
Every offer on the live market comes from a reviewed supplier and is sold by EverXCloud Inc.