China just dropped a new frontier model... and nobody can explain how
Kimi K3.1's frontier performance contradicts Anthropic's claim of blocking 90-99% of Chinese distillation attacks
“To them, it's grand theft autocomplete.”
Moonshot AI released Kimi K3.1, a 3-trillion-parameter open-weight model that first appeared anonymously on Chatbot Arena as 'Kibitz' before its official reveal, scoring within one point of Opus 5.5 on Terminal Bench 4.0. The launch creates a credibility puzzle: Anthropic recently claimed to have blocked 90-99% of Chinese distillation attacks, yet Kimi K3.1's performance benchmarks are arguably impossible if that claim holds. This tension forces a reassessment of whether Chinese labs are genuinely closing the frontier gap through independent research rather than distillation.