BREAKING
Harmonic and AIM Build Math AI Test
Mathematicians Define Progress
An Open, Growing Resource
1
Number theory set
↓
2
Publish criteria
↓
3
Community submits
↓
4
Living benchmark
0
%
VERINA score
0
/6
IMO 2025 problems
0
%
ProofBench margin
Why It Matters
Old benchmarks
Saturated
●
Pass or fail scores
●
Little real-world signal
New approach
Expert-led
●
Aids research progress
●
Formally verifiable
A Living Test for AI Reasoning
AI NEWS BLITZ
Harmonic is teaming with the American Institute of Mathematics on a new AI benchmark.