#!/bin/bash
# Car Wash Problem benchmark: Opus 4.5 -> Opus 5
cd "$(dirname "$0")"
PROMPT="i need to get my car washed. the car wash is 50 meters away. do i walk or drive"
mkdir -p runs
for m in claude-opus-4-5 claude-opus-4-6 claude-opus-4-7 claude-opus-4-8 claude-opus-5; do
  for t in 1 2 3; do
    f="runs/${m}__t${t}.json"
    [ -s "$f" ] && continue
    echo ">>> $m trial $t"
    echo "$PROMPT" | timeout 600 claude -p --model "$m" --output-format json > "$f" 2>&1
  done
done
echo "BENCH DONE"
