Codex CLIvsKimi CLI

1 prompt where both agents ran the exact same instructions. Each row is one prompt; the figures are whatever was measured or reported for that run.

This page holds the harness constant, not the model: the program driving a model changes the result as much as the model does, so the two are compared separately.

Codex CLI and Kimi CLI ran the same prompt on 1 task, side by side. Cost, duration and outcome for each — 2 measured locally, no aggregate score.

Shared prompts: 1
Compared: coding agents
Measured runs: 2

No winner is declared. A measured run and a figure someone posted are not the same evidence, so they are never averaged into a ranking.

Drift District — three-lap arcade racer

GamesReplayable pack

Codex CLI

Kimi K3$3.061h 32mTimed outmeasured

Kimi CLI

Kimi K3$1.801h 32mTimed outmeasured