According to LessWrong, open-source AI models consistently lag behind proprietary frontier models by several months. Researcher RobinHa used Epoch's ECI metric—which evaluates capability across multiple benchmarks—to measure this gap more rigorously than industry indices. The analysis reveals a persistent pattern: as companies like OpenAI and Anthropic advance the frontier, open-weight releases remain several months behind, with the gap tracked down to the month.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton