Kimi K3 took first on SpreadsheetBench 2, surpassing Claude Fable 5 to become the first open-weight model to beat every closed rival on a benchmark. One observer called the earlier Frontend Code Arena win a DeepSeek 2.0 moment.
Four launches in eight days lifted six labs above 50 on the intelligence index, up from two in June. The top three models now span three points across three labs.