⌘K

Compare GPUs for Local AI

Find the right GPU for LLMs, image generation, video generation, fine-tuning, and more. Compare VRAM, benchmarks, software compatibility, and value — then buy with confidence.

64 GPUs tracked LLM · Image · Video · Training Independent · Data-sourced
CompareAIHardware.com earns from qualifying purchases via Amazon Associates. See our affiliate disclosure. Prices and availability not displayed — always verify on Amazon.

GPU Comparison Selector

Add 2-4 GPUs to compare

What will you use it for?

All GPUs (64)

Click a row to add to comparison, or use the selector above.

GPUMfrVRAMBWTDPMSRPPriceYearStatus
Arc B570 Intel 10 GB 380 GB/s 150W $219 Check Price 2025 available
NVIDIA RTX PRO 6000 Blackwell NVIDIA 96 GB 1792 GB/s 600W Check Price 2025 available
NVIDIA B200 NVIDIA 192 GB 8000 GB/s 1000W 2025 available
M3 Ultra (Mac Studio) Apple 512 GB 819 GB/s 270W $3,999 Check Price 2025 available
NVIDIA RTX PRO 5000 Blackwell NVIDIA 48 GB 1344 GB/s 300W 2025 available
GeForce RTX 5070 Ti NVIDIA 16 GB 896 GB/s 300W $749 Check Price 2025 available
NVIDIA RTX PRO 4500 Blackwell NVIDIA 32 GB 896 GB/s 200W Check Price 2025 available
NVIDIA RTX PRO 4000 Blackwell NVIDIA 24 GB 672 GB/s 145W Check Price 2025 available
GeForce RTX 5070 NVIDIA 12 GB 672 GB/s 250W $549 Check Price 2025 available
GeForce RTX 5060 Ti 16GB NVIDIA 16 GB 448 GB/s 180W $429 Check Price 2025 available
NVIDIA RTX PRO 2000 Blackwell NVIDIA 16 GB 288 GB/s 70W Check Price 2025 available
GeForce RTX 5060 NVIDIA 8 GB 448 GB/s 145W $299 Check Price 2025 available
Radeon RX 9070 XT AMD 16 GB 640 GB/s 304W $599 Check Price 2025 available
GeForce RTX 5090 NVIDIA 32 GB 1792 GB/s 575W $1,999 Check Price 2025 available
GeForce RTX 5050 NVIDIA 8 GB 320 GB/s 130W $249 Check Price 2025 available
Radeon RX 9070 AMD 16 GB 640 GB/s 220W $549 Check Price 2025 available
GeForce RTX 5080 NVIDIA 16 GB 960 GB/s 360W $999 Check Price 2025 available
NVIDIA B300 (Blackwell Ultra) NVIDIA 288 GB 8000 GB/s 1400W 2025 available
AMD Instinct MI355X AMD 288 GB 8000 GB/s 1400W 2025 available
Apple M4 Apple 32 GB 120 GB/s 2024 available
Apple M4 Pro Apple 64 GB 273 GB/s 2024 available
Radeon RX 7600 XT AMD 16 GB 288 GB/s 190W $329 Check Price 2024 available
Arc B580 Intel 12 GB 456 GB/s 190W $249 Check Price 2024 available
H200 SXM NVIDIA 141 GB 4800 GB/s 700W 2024 available
AMD Instinct MI325X AMD 256 GB 6000 GB/s 1000W 2024 available
GeForce RTX 4070 Ti SUPER NVIDIA 16 GB 672 GB/s 285W $799 Check Price 2024 eol
GeForce RTX 4080 SUPER NVIDIA 16 GB 736 GB/s 320W $999 Check Price 2024 available
GeForce RTX 4070 SUPER NVIDIA 12 GB 504 GB/s 220W $599 Check Price 2024 eol
M4 Max (MacBook Pro) Apple 128 GB 546 GB/s $3,199 2024 available
GeForce RTX 4060 Ti 16GB NVIDIA 16 GB 288 GB/s 160W $499 Check Price 2023 available
GeForce RTX 4060 NVIDIA 8 GB 272 GB/s 115W $299 Check Price 2023 available
Apple M3 Apple 24 GB 150 GB/s 2023 available
Apple M3 Pro Apple 36 GB 300 GB/s 2023 available
H100 SXM NVIDIA 80 GB 3350 GB/s 700W 2023 available
Arc A580 Intel 8 GB 512 GB/s 185W $179 Check Price 2023 available
Apple M3 Max Apple 128 GB 400 GB/s 2023 available
RTX 6000 Ada NVIDIA 48 GB 960 GB/s 300W $6,800 Check Price 2023 available
NVIDIA RTX 5000 Ada NVIDIA 32 GB 576 GB/s 250W $2,250 Check Price 2023 available
GeForce RTX 4060 Ti 8GB NVIDIA 8 GB 288 GB/s 160W $399 Check Price 2023 available
NVIDIA H100 PCIe 80GB NVIDIA 80 GB 2039 GB/s 350W 2023 available
AMD Instinct MI300X AMD 192 GB 5300 GB/s 750W 2023 available
Radeon RX 7800 XT AMD 16 GB 624 GB/s 263W $499 Check Price 2023 available
Radeon RX 7700 XT AMD 12 GB 432 GB/s 245W $419 Check Price 2023 available
GeForce RTX 4070 NVIDIA 12 GB 504 GB/s 200W $549 Check Price 2023 eol
Radeon RX 7600 AMD 8 GB 288 GB/s 165W $269 Check Price 2023 available
GeForce RTX 3090 Ti NVIDIA 24 GB 1008 GB/s 450W $1,999 Check Price 2022 eol
Radeon RX 7900 XTX AMD 24 GB 960 GB/s 355W $999 Check Price 2022 available
Arc A770 16GB Intel 16 GB 560 GB/s 225W $349 Check Price 2022 available
Arc A750 Intel 8 GB 512 GB/s 225W $289 Check Price 2022 available
AMD Radeon Pro W7900 AMD 48 GB 864 GB/s 295W $3,999 Check Price 2022 available
AMD Radeon Pro W7800 AMD 32 GB 576 GB/s 260W $2,499 Check Price 2022 available
GeForce RTX 4080 NVIDIA 16 GB 716.8 GB/s 320W $1,199 Check Price 2022 eol
Radeon RX 7900 XT AMD 20 GB 800 GB/s 315W $649 Check Price 2022 available
GeForce RTX 4090 NVIDIA 24 GB 1008 GB/s 450W $1,599 Check Price 2022 available
GeForce RTX 3080 Ti NVIDIA 12 GB 912 GB/s 350W $1,199 Check Price 2021 eol
GeForce RTX 3060 12GB NVIDIA 12 GB 360 GB/s 170W $329 Check Price 2021 available
RTX A6000 NVIDIA 48 GB 768 GB/s 300W $4,500 Check Price 2021 available
NVIDIA A100 80GB SXM NVIDIA 80 GB 2039 GB/s 400W 2021 available
Radeon RX 6800 XT AMD 16 GB 512 GB/s 300W $649 Check Price 2020 eol
GeForce RTX 3080 10GB NVIDIA 10 GB 760 GB/s 320W $699 Check Price 2020 eol
GeForce RTX 3070 NVIDIA 8 GB 448 GB/s 220W $499 Check Price 2020 eol
NVIDIA A100 40GB SXM NVIDIA 40 GB 1555 GB/s 400W 2020 eol
GeForce RTX 3090 NVIDIA 24 GB 936 GB/s 350W $1,499 Check Price 2020 eol
GeForce RTX 2080 Ti NVIDIA 11 GB 616 GB/s 250W $999 Check Price 2018 eol

💾 Why VRAM matters most

For local AI, VRAM determines which models you can run at all. A GPU with 8GB simply cannot load a 14B model, regardless of speed. Bandwidth then determines how fast inference runs. Prioritize VRAM capacity first, then bandwidth.

⚡ Measured vs theoretical

TFLOPS numbers from spec sheets are theoretical. Actual LLM inference depends on memory bandwidth, quantization support, software optimization, and thermal behavior. We label every data point: Measured, Vendor spec, Calculated, Community, or Estimated.

🔧 Software ecosystem

NVIDIA (CUDA) has the broadest support: PyTorch, vLLM, Ollama, ComfyUI, TensorRT all work out of the box. AMD (ROCm) works on Linux for most tools. Apple (MLX) is excellent for LLMs but limited for some video/image tools. Intel is improving but has less community support.

💰 New vs used considerations

Used RTX 3090 (24GB) cards offer exceptional value for LLM work. Check: memory condition (mining wear), fan health, thermal pad condition, and warranty status. Verify VRAM capacity matches (some models have variants).

🔗 Multi-GPU limitations

Most local AI tools split models across GPUs via tensor parallelism or pipeline parallelism. Performance scales well for inference but NOT linearly. NVLink helps but is absent on consumer RTX 40/50 series. Expect ~80% scaling with 2 GPUs for inference.

🏆 Best GPUs for LLM inference

Full tier-by-tier ranking of every current GPU for local LLMs — Best GPUs for LLM Inference.

❓ How to choose

1. What\\'s your largest model? → Minimum VRAM
2. What\\'s your workload? → LLM/Image/Video
3. What\\'s your software preference? → CUDA/ROCm/MLX
4. What\\'s your budget? → New + used options
5. What\\'s your PSU/case? → Physical fit