Will Bryk
@WilliamBryk
ATLAS is the hardest search eval in the world.
Existing search evals are too easy for frontier models (saturated) and have already been trained on (contaminated).
We created hundreds of real-world valuable search tasks that frontier models score <50%
To keep it fresh and
Existing search evals are too easy for frontier models (saturated) and have already been trained on (contaminated).
We created hundreds of real-world valuable search tasks that frontier models score <50%
To keep it fresh and
Exa@ExaAILabs · Oct 8We built ATLAS, a web search benchmark for how agents actually search today.
Existing agentic search benchmarks are often outdated, contaminated, or saturated. ATLAS is designed to separate strong search from weak search.
Learn more: exa.ai/blog/atlas-ben…
Existing agentic search benchmarks are often outdated, contaminated, or saturated. ATLAS is designed to separate strong search from weak search.
Learn more: exa.ai/blog/atlas-ben…
Open quoted post →
11 70