Kimi K3 Tops Open-Source ARC-AGI-2 Leaderboard, Rivaling Closed-Source Models

Moonshot's open-source Kimi K3 model achieved a breakthrough score of 60.4% on the ARC-AGI-2 benchmark, topping the open-source leaderboard. Verified by ARC Prize, the model demonstrates exceptional abstract reasoning, rivaling closed-source frontiers like Claude Opus at high cost-efficiency.

2026-08-01 ~ 2026-08-01 · 3 related posts

• Kimi-K3 Scores 60.4% on ARC-AGI-2 Benchmark — scaling01 · 2026-08-01
• Kimi K3 Tops Open-Weight Models on ARC-AGI, Rivaling Frontier Models — mhmazur · 2026-08-01

1 near-duplicate retellings: mhmazur