Start with the rig
Capacity, memory type, software stack, and realistic caveats for local AI.
NVIDIA hardware
RTX 3060 12GB
budget used GPU with useful 12 GB capacity
12 GB VRAMInspect →CUDARTX 3080 10GB
fast older GPU constrained by 10 GB VRAM
10 GB VRAMInspect →CUDARTX 3090 24GB
used-market 24 GB local AI favorite
24 GB VRAMInspect →CUDARTX 4060 Ti 16GB
efficient 16 GB consumer option
16 GB VRAMInspect →CUDARTX 4070 Ti Super 16GB
faster 16 GB card for mid-size models
16 GB VRAMInspect →CUDARTX 4080 Super 16GB
high throughput but still limited to 16 GB
16 GB VRAMInspect →CUDARTX 4090 24GB
strong 24 GB inference and image generation
24 GB VRAMInspect →CUDARTX 5060 Ti 16GB
current mid-range 16 GB option
16 GB VRAMInspect →CUDARTX 5070 Ti 16GB
current performance card with 16 GB ceiling
16 GB VRAMInspect →CUDARTX 5080 16GB
fast compute with a 16 GB memory ceiling
16 GB VRAMInspect →CUDARTX 5090 32GB
fast 32 GB consumer flagship
32 GB VRAMInspect →CUDARTX A6000 48GB
workstation capacity for larger quantized models
48 GB VRAMInspect →CUDARTX PRO 6000 Blackwell 96GB
large-memory professional workstation GPU
96 GB VRAMInspect →CUDAL40S 48GB
data-center inference and image generation
48 GB VRAMInspect →CUDAA100 80GB
large-memory data-center accelerator
80 GB VRAMInspect →CUDAH100 80GB
high-end production accelerator
80 GB VRAMInspect →CUDADGX Spark 128GB
desktop AI system with unified memory
128 GB UnifiedInspect →AMD hardware
Radeon RX 7900 XTX 24GB
24 GB value with a more selective software path
24 GB VRAMInspect →ROCmRadeon PRO W7900 48GB
large VRAM; verify framework and OS compatibility
48 GB VRAMInspect →ROCm / VulkanFramework Desktop Ryzen AI Max+ 128GB
compact unified-memory PC; software support matters
128 GB UnifiedInspect →