AI Benchmark
Kimi K3.1 Pro: Coding Benchmarks and Safety Trade-offs
Moonshot’s Kimi K3.1 Pro beats GPT-6 Astra in coding but deliberately reduced its cyber-offensive capability from 90.5% to 81.0%. Full benchmark.
Explore all published articles in this archive.
Moonshot’s Kimi K3.1 Pro beats GPT-6 Astra in coding but deliberately reduced its cyber-offensive capability from 90.5% to 81.0%. Full benchmark.
Security firm Mindgard says Moonshot’ Kimi K2.6-K3 Swarm gave bioweapon & assassination instructions after a jailbreak. Full breakdown of what we know
Moonshot AI announced Kimi K3, China’s challenger to GPT and Claude. Learn about its capabilities.
Liang Wenfeng, Yang Zhilin and Tang Jie are associated with three different Chinese AI companies: DeepSeek, Moonshot AI and Zhipu/Z.ai. Their backgrounds help explain the organizations’ origins, but a founder profile is not a benchmark of model quality.
