[Third-Party Evaluation: Kimi K3 Agent Tasks Approach Fable 5, but Hallucinations Increase] According to monitoring by Beating Insights, Artificial Analysis has released an independent evaluation of Kimi K3. It scored 57 on the Intelligence Index v4.1. Its overall performance is comparable to Claude Opus 4.8 and GPT-5.5 but remains below Claude Fable 5 and GPT-5.6 Sol. This comprehensive index consists of 9 tests, covering real-world tasks, tool utilization, programming, reasoning, knowledge, and long-context processing. Kimi K3's Agent performance stands out. It achieved 1668 Elo in the real-world task test GDPval-AA v2, surpassing Opus 4.8, GPT-5.5, and GLM-5.2. In automated SaaS workflow testing, it scored 53%, ranking first. For long-process knowledge work, it ranked second, trailing only Fable 5. Within this index, Kimi K3's average cost per task is $0.94. GPT-5.6 Sol costs $1.04, while Opus 4.8 costs $1.80. However, its hallucination rate in knowledge reliability tests increased from 39% to 51%, while accuracy improved from 33% to 46%. [Original Link]