—
Simple — 0/25 raced
Keep going
Racing Haiku 4.5 (Anthropic) against a typical person on outline a financial model: 3 hr by hand versus 8.3 s for the model, 1,309× faster, 57% accurate on this task's rubric. ✗ Too small for this one. Try Opus 4.8. It used 430 tokens for a cost of $0.0016 — see how that stacks up against every other model →
Choose a task you do every week, then pick an AI model. Watch how fast it finishes, and see what happened in those seconds: what it read, understood, thought through, wrote — and how accurate it was.
Run the same task on Claude and GPT side by side — with effort levels and Extended Thinking.
"You" is a typical time for a person doing the task by hand. "AI" is a recorded run for the selected model, shown to scale. Stage times, accuracy, and token cost are placeholder profiles until runs are recorded and scored. How we measure → · FAQ → · Model comparison →