quanteval.ai tracks how quantized open-source models perform on coding, reasoning, and tool-use work. Reports explain the benchmark results; leaderboards, speed tests, and prompt tests keep the raw comparisons available.