Retired benchmark. These historical results are preserved for inspection and excluded from active-suite run counts. Browse the archive →

Focused model compare

Universe Simulator

Grok 4.6 is the starting run, compared with GPT-6 Astra.