TeleAI Unveils Breakthrough Metric to Quantify AI “Talent” in Large Language Models

Beijing, Dec. 19, 2025 (GLOBE NEWSWIRE) -- In a major advancement for AI model evaluation, the Institute of Artificial Intelligence of China Telecom (TeleAI) has introduced a groundbreaking metric—Information Capacity—that redefines how large language models (LLMs) are assessed beyond traditional size-based comparisons. This new approach reveals that true model “talent” lies not in size, but in how efficiently a model compresses and processes knowledge relative to its computational cost. Information capaci ...