★ Top story · ModelsJul 16
Kimi's open model K3 nears GPT-5.6 Sol and Fable 5 while signaling the end of super cheap Chinese AI
Moonshot released Kimi K3, a large mixture-of-experts model that posts frontier-level benchmark scores, including an Epoch Capabilities Index of 155.6, 93% on GPQA Diamond, and 97% on AIME. Early testers say it beats Fable 5 on some frontend coding tasks while trailing on others. It launched API-only at $3 per million input tokens and $15 per million output, well above Moonshot earlier Kimi models, with open weights committed for July 27.