GPU Models for LLM Inference — Specs, VRAM & Prices
Browse and compare GPU hardware for LLM inference
GeForce RTX 5090
NVIDIA · Blackwell
VRAM
32 GB
Memory BW
1,792 GB/s
FP16
419.2 TFLOPS
MSRP
USD 1,999.00
GeForce RTX 3090
NVIDIA · Ampere
VRAM
24 GB
Memory BW
936 GB/s
FP16
142 TFLOPS
MSRP
USD 1,499.00
GeForce RTX 4090 D
NVIDIA · Ada Lovelace
VRAM
24 GB
Memory BW
1,008 GB/s
FP16
290.6 TFLOPS
MSRP
USD 1,835.00
AMD Radeon RX 7900 XTX
AMD · RDNA3
VRAM
24 GB
Memory BW
960 GB/s
FP16
61.44 TFLOPS
MSRP
USD 999.00
Apple M2 Ultra
Apple · M2
VRAM
64 GB
Memory BW
800 GB/s
FP16
54.4 TFLOPS
MSRP
USD 3,999.00
Apple M2 Max
Apple · M2
VRAM
32 GB
Memory BW
400 GB/s
FP16
27.2 TFLOPS
MSRP
USD 1,999.00
Apple M4 Pro
Apple · M4
VRAM
24 GB
Memory BW
273 GB/s
FP16
17 TFLOPS
MSRP
USD 1,399.00
Apple M4 Max
Apple · M4
VRAM
36 GB
Memory BW
546 GB/s
FP16
36.8 TFLOPS
MSRP
USD 1,999.00
Apple M3 Max
Apple · M3
VRAM
36 GB
Memory BW
400 GB/s
FP16
28.2 TFLOPS
MSRP
USD 3,199.00
GeForce RTX 3090 Ti
NVIDIA · Ampere
VRAM
24 GB
Memory BW
1,008 GB/s
FP16
160 TFLOPS
MSRP
USD 1,999.00
Apple M1 Max
Apple · M1
VRAM
32 GB
Memory BW
400 GB/s
FP16
20.8 TFLOPS
MSRP
USD 1,999.00
Apple M1 Ultra
Apple · M1
VRAM
64 GB
Memory BW
800 GB/s
FP16
42 TFLOPS
MSRP
USD 3,999.00
GeForce RTX 4090
NVIDIA · Ada Lovelace
VRAM
24 GB
Memory BW
1,008 GB/s
FP16
330.3 TFLOPS
MSRP
USD 1,599.00