This selection tool helps you find a suitable GPU for AI or deep-learning projects. It considers your budget, whether you're an individual or organization, and your use case (training large models, inference, or creative workloads). Recommendations cover NVIDIA Blackwell, Ada, and Ampere generations, plus practical cloud and used-market options.
Explore non-GPU AI accelerators like ASICs, FPGAs, and NPUs, comparing them to GPUs for training and inference. Includes major examples, performance trends, and cloud access options.
This calculator estimates the GPU memory needed to run LLM inference. Select the model size and precision (FP32 - FP4) to get a quick memory range estimate.