anykey.ai

Quantization Benchmark

Run the same model in float32, float16, and int8 to compare inference speed, memory footprint, and prediction stability.

How it works

input image

No image selected yet.

Pick a preset or upload an image, then run the benchmark.

speed + memory

Run the benchmark to visualize speed and memory differences.

quality comparison

Quality metrics will appear after the benchmark runs.

weight inspector

Run the benchmark to inspect how values are rounded per precision.