Artificial Analysis has released six new industry-specific performance indices for AI models, with Anthropic's Claude Fable 5 leading every category. The benchmarks cover domains like finance, law, and healthcare, based on US O*NET occupational classifications. The results show Fable 5 outperformed other models, including OpenAI's GPT-5.5 and Google's Gemini 3.5 Flash, in all six indices. The benchmarks are weighted by the frequency of specific skills in each industry, with all tests conducted independently. Source: thedecoder
The performance of Claude Fable 5 was consistent across all eight indices, including the new ones from Artificial Analysis. It was followed by Claude Opus 4.8 and GPT-5.5 in most categories, with GPT-5.6 potentially closing the gap. In the Engineering benchmark, GLM-5.2 (max) scored 53 points, placing it fifth, just two points behind GPT-5.5 (xhigh) and Claude Sonnet 5 (max). OpenAI's GPT-5.5 (xhigh) also led in the Strategy & Ops Index with 55 points, while DeepSeek V4 Pro (max) took the open-weights lead with 38 points. Source: thedecoder
The cost of using Claude Fable 5 is significantly higher than cheaper alternatives. While DeepSeek V4 Flash (max) handles tasks for less than $0.04 per task, Fable 5 costs $3.48 per task in the Strategy & Ops Index, over 100 times more. This premium is debated among enterprises, with some suggesting pairing models to optimize cost-effectiveness. Ethan Mollick, a US economist, argues that current benchmarks may not fully capture Fable 5's performance, but there is no evidence to support this claim. Source: thedecoder
