Our Benchmarks — evaluations The Aggregate runs itself
Evaluations we run ourselves, from recall and reasoning to predicting model performance. Choose a benchmark to explore its results and method.
Guesswork
Predicting benchmark scores before they exist. Live, updated daily.
Long-tail book recall
Closed-book questions about details in 60 public-domain books, from the famous to the obscure.
Exotic fact recall
Closed-book lookups from five public registries: proteins, ISBNs, museum objects, asteroids and network ports.
Asteroid numbers
Name 48 freshly sampled asteroids' catalogue numbers, from memory.
Day-one recall check
The 34-question recall test every new free model gets on the day it appears.
Forecasting resolved questions
Probabilities for 76 prediction-market questions that no model in the panel could look up from memory.
Inversion and execution ladder
24 generated puzzles: trace a short program, or undo a hidden digit transform, at rising difficulty.
Puzzle portfolio ladder
30 generated puzzles in five families: logic grids, knapsacks, document facts, specifications and bug hunts.
Guess the chess rating
Estimate each player's established Lichess rapid rating from an anonymized game.
Which chess player is stronger?
Compare two anonymized games and choose the target player with the higher established Lichess rapid rating.
What happened on the trip?
Guess what really happened on a two-day trip from its plan alone.
Can it write like me?
Turn neutral drafts of the site author's real messages back into the author's own voice.
Also run here: smaller experiments, single-model runs and qualification checks.
Interactive version: theaggregate.ai/our-benchmarks · How It Works · Data refreshed daily, snapshot 2026-11-06.