Kimi K3.1 Pro is presented as a competitive open-weight model, scoring strongly across several benchmarks; the charts show it outperforming GPT-6 Astra and Fable 5.1 on some coding and task benchmarks, while trailing on others. The Terminal-Bench chart shows Kimi K3.1 Pro scoring about 66% at roughly $5 per task, versus higher costs for the shown Claude models. The post also questions what its rise could mean for Anthropic and infrastructure being built for it.