Woofun AI data shows that four new models—Grok 4.5, GPT-5.6, Muse Spark 1.1, and Kimi K3—entered the top tier within eight days. The number of teams exceeding an Intelligence Quotient of 50 rose to six, up from two in early June.

Kimi K3 achieved a score of 57, ranking third behind Claude Fable 5 (60) and GPT-5.6 Sol (59), while surpassing Claude Opus 4.8 (56). Cost per task for Kimi K3 is $0.94, approximately half that of Opus 4.8. GPT-5.6 Sol costs $1.04, whereas Claude Fable 5 requires $2.75. Grok 4.5 scored 54 at $0.31 per task, reducing cutting-edge model costs to one-third of previous levels.