AIAstra reportedly leads a 120-match DOOM benchmark with an 82.5% win rate
A user says their GPT-6 agent benchmark found different leaders for speed and value: Sol averaged 5.34 seconds per decision, while Luna had the most wins per dollar at 12.8.
TLDR
A user reports benchmarking GPT-6 models by having agents fight each other across 120 DOOM matches. In their reported results, Astra had the highest win rate at 82.5%, Sol made the fastest decisions at an average of 5.34 seconds, and Luna had the most wins per dollar at 12.8. They also shared a GitHub repository for the agent arena.
Astra reportedly leads a 120-match DOOM benchmark with an 82.5% win rate
A user says their GPT-6 agent benchmark found different leaders for speed and value: Sol averaged 5.34 seconds per decision, while Luna had the most wins per dollar at 12.8.
