L40S
Best local AI models for this GPU
VRAM
48 GB
Bandwidth
864 GB/s
Vendor
NVIDIA
Compatible Models
28
Where to Buy
Top Compatible Models
Top 10 models ranked by performance. Click any model to expand details or view full compatibility.
Memory Usage44.3 GB
Weights
KV Cache
Overhead
Context Length163,840 tokens
Install with Ollama:
ollama run deepseek-r1:70b
Memory Usage44.3 GB
Weights
KV Cache
Overhead
Context Length131,072 tokens
Install with Ollama:
ollama run llama3.3:70b
Memory Usage45.5 GB
Weights
KV Cache
Overhead
Context Length131,072 tokens
Install with Ollama:
ollama run qwen2.5:72b
Memory Usage44.3 GB
Weights
KV Cache
Overhead
Context Length131,072 tokens
Install with Ollama:
ollama run llama3.1:70b
Memory Usage24.3 GB
Weights
KV Cache
Overhead
Context Length131,072 tokens
Install with Ollama:
ollama run qwen3:32b-q5_k_m
Memory Usage25.3 GB
Weights
KV Cache
Overhead
Context Length16,384 tokens
Install with Ollama:
ollama run gemma3:27b-q5_k_m
Memory Usage20.1 GB
Weights
KV Cache
Overhead
Context Length131,072 tokens
Install with Ollama:
ollama run mistral-small:24b-q5_k_m
Memory Usage24.3 GB
Weights
KV Cache
Overhead
Context Length131,072 tokens
Install with Ollama:
ollama run qwen2.5:32b-q5_k_m
Memory Usage24.3 GB
Weights
KV Cache
Overhead
Context Length131,072 tokens
Install with Ollama:
ollama run qwen2.5-coder:32b-q5_k_m
Memory Usage24.9 GB
Weights
KV Cache
Overhead
Context Length131,072 tokens
Install with Ollama:
ollama run qwen3:30b-a3b-q5_k_m