Sparring

Model reviews and benchmarks, scored on real work.

On every frontier release

1 piece

A Two-Point Lead Is Noise Until Proven Otherwise

Coding-agent leaderboards report decimals. The VM, time limit, harness, and leaked answer keys can move the score more than the model.

Tim CheungAug 26, 20265 min read

Follow Sparring

We put frontier models through the same fixed gauntlet: long-context reading, refactors, research, and taste. No vendor decks, no cherry-picked demos.