Find GPU for Your Model

Running Kimi K2.6? DeepSeek V4 Pro? Mercury 2? Or a custom model? Pick the parameter count and precision, and we show every GPU in our catalog that has enough VRAM — ranked by FP16 compute performance.

We calculate VRAM requirements for FP16, INT8, and FP4/GGUF quantization across batch sizes. Each result shows the maximum batch size that GPU can handle for your config.