[Kimi K3 White-Collar Tasks Approach Fable 5, Costs Rise to 10x the Previous Generation] According to monitoring by Beating Insights, Artificial Analysis has updated the AA-Briefcase rankings. This evaluation requires models to extract information from nearly 2,000 emails, Slack records, and company documents, then deliver spreadsheets, presentations, and interface prototypes. It includes 4 long-term projects and 91 confidential tasks. Kimi K3 scored 1543 Elo, second only to Claude Fable 5's 1574. It surpassed GPT-5.6 Sol's 1501 and also outperformed Claude Sonnet 5 and Claude Opus 4.8. K3's objective task pass rate was 51%, compared to Fable 5's 56%. However, K3's analysis quality score was slightly higher, at 1754 versus 1744. Its main shortfall was in final product presentation, where it lagged behind GPT-5.6 Sol and Opus 4.8. The performance improvements came with significantly higher token consumption. K3's average cost per task was $10.57, approximately 10 times that of Kimi K2.6. On average, it executed 83 rounds, output 120,000 tokens, and took 56.4 minutes. The time required to complete similar tasks was about 2.5 times that of Fable 5. [Original Link]