Benchmark
GDPval-AA
A benchmark for evaluating the performance of AI on GDP-related tasks, using an Elo rating system.
Timeline (last 90 days)
Oct 18
GDPval-AA is an evaluation framework oriented toward economically valuable work output, where AA usually stands for Agentic Assessment and v2.1 is an iterative version of it.
Unverified50%
Oct 16
On the GDPval-AA benchmark for human preference evaluation, Claude Opus 4.6 scored 1606 Elo and Gemini 3.1 Pro scored 1317 Elo.
Unverified50%
All Facts (2)
Unverified
GDPval-AA is an evaluation framework oriented toward economically valuable work output, where AA usually stands for Agentic Assessment and v2.1 is an iterative version of it.
50%UnverifiedOn the GDPval-AA benchmark for human preference evaluation, Claude Opus 4.6 scored 1606 Elo and Gemini 3.1 Pro scored 1317 Elo.
50%