Qwen 3.8 27B achieves surprisingly high 52-point score on the Artificial Analysis Intelligence Index

The focus is the Qwen 3.8 27B language model, which scored 52 points on the Artificial Analysis Intelligence Index. The score is mentioned in a report shared on Hacker News and indicates that Qwen 3.8 27B's performance matches that of the GPT-5.6 Luna (max) model. Qwen 3.8 27B trails by one point behind the GLM-5.2 (max) and DeepSeek V4 Pro 0813 (max) models, which also showed high performance. GLM-5.2 has 753 billion parameters, while DeepSeek V4 Pro 0813 has only 1.6 billion parameters; GPT-5.6 Luna's size was not disclosed but is likely much larger than 27 billion. Importantly, Qwen 3.8 27B is relatively small (27 billion parameters) yet produced a competitive result, which could influence debates about model cost-effectiveness and scalability. The performance differences highlight that parameter count alone is not always determinative of capabilities. The report's details do not include an exact time for the measurement, but the result has attracted attention from the professional community. A possible consequence is increased interest in smaller models and investment in improved efficiency optimizations.