Side by side

brAIAIn vs. the answer key.

Eight real problems from GSM8K, the grade-school math test frontier labs report scores on. The correct answer is from the test. brAIAIn's answer is generated live, in your browser, by the current model. We wrote none of the answers.

Best published score-

brAIAIn's score-

Model answering below-

brAIAIn has seen none of these problems. It has only read these sentences. Its full GSM8K run, all 1,319 problems, is on Benchmarks.