LiveBench Overall

In plain words · Overall score on questions refreshed monthly so they cannot be memorized

What does it measure?

An overall score from an exam that is freshly written every month. It spans every subject — reasoning, coding, agentic coding, maths, data analysis, language, instruction following — so it shows whether an AI is well-rounded rather than strong in one subject.

Who evaluates it, and how?

LiveBench, built by researchers at Abacus.AI, NYU and elsewhere, releases new questions every month from recent competitions, papers and news, reducing the chance that an AI has seen the questions during training (contamination). Only questions with fixed answers are used, graded by program rather than by another AI. This table takes the latest monthly results published at livebench.ai; an AI missing from the newest release is listed separately below the table with its last published score, unranked. A subject score is the average of its tasks. The overall score in this table is computed as the average of every task score we import (agentic coding included), so it can differ slightly from the six-subject average shown on the LiveBench site.

Reading the score

0–100. The questions are designed to be hard — even top AIs struggle to exceed 70 — and because they change monthly, avoid comparing scores from different dates.

Last updated: 2026-06-25