Same four targets, same prompt, same Bash tool. GPT-6 Sol verified 6 of 11 runs in 1h 06m for $25.38 (est.). Grok 4.7 verified 3 of 11 in 4h 53m for $191.31, and issued 2,929 Bash commands, the most of any model on the board.
enclave.ai/hackingrace