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

Focused model compare

Low-Poly World Explorer

Claude Fable 5.1 is the starting run, compared with GPT-6 Astra, Grok 4.7.