METR says Anthropic’s AI models are improving at roughly twice the historical pace

Anthropic’s AI models are advancing at roughly twice the historical rate, according to an assessment by METR, a finding that signals faster-than-expected capability gains and heightens both opportunity and safety concerns. The evaluation points to a doubling of the pace of improvement relative to past benchmarks, though the article did not disclose specific model versions, benchmark scores, or token prices. The result matters because it suggests AI progress may outrun the assumptions behind current safety and deployment planning. What to watch next is whether subsequent METR evaluations confirm the accelerated trend and how Anthropic adjusts its safety and release cadence in response.