Kimi K3 Tops the Writing Leaderboard

oran_ge · x · 2026-07-17

An account citing internal writing benchmark results claims that Kimi K3 has climbed to #1 on their editorial-style writing leaderboard, reaching an Elo of 2840 and surpassing Claude Fable 5.

More impressively, it jumped from #21 to #1 compared to its predecessor, Kimi K2.6. The post also mentions that K3's scripting cost per piece is about $0.25, which is roughly 5 times cheaper than the model it dethroned.

The core takeaway of this information isn't just "another model release," but rather the magnitude of its leap on a specific writing benchmark and its significantly lower inference costs.

Related event: Kimi K3 Triggers a Reassessment of Chinese Frontier AI(94 posts)→

Original post →

More from Models

Models channel →