Claude's Performance Gains Exceed METR's Projections

According to METR(Model Evaluation and Threat Research)'s 'Time Horizons' analysis, the pace of recent AI models' capability development is significantly exceeding the existing predicted curve.

In particular, the performance improvement of the Claude model shows that AI's autonomous problem-solving ability is developing much faster than expected. This is leading to an earlier arrival at the technical threshold points that the existing METR graph had predicted.

This change suggests that AI's Capability growth rate is accelerating non-linearly, and it is being evaluated as a very important indicator for establishing AI safety and regulatory response strategies.