Anthropic Launches Claude Sonnet 5 With Improved Agentic Performance; US Lifts Export Controls on Advanced Models

Anthropic released Claude Sonnet 5, claiming enhanced agentic search capabilities over Sonnet 4.6 while matching Opus 4.8 performance at lower cost, with a 63.2% score on agentic coding benchmarks. The company also announced that the US government has removed export restrictions on its Claude Mythos 5 and Fable 5 models, reversing earlier controls imposed on the cybersecurity-trained systems.

Why it matters: The release demonstrates competitive progress in AI agent capabilities while the policy reversal signals shifting US government priorities on AI export controls, directly affecting Anthropic's ability to serve international markets.