Anthropic's Sonnet 5.5 rises in the rankings

Anthropic has released Claude Sonnet 5.5, its latest midtier model, as competition in the AI market intensifies ahead of OpenAI’s DevDay. The company says the new version is faster, cheaper to run, and closes much of the gap with its higherend Opus model on some knowledgework and coding tasks. According to the newsletter, Sonnet 5.5 keeps the same pricing as before while improving performance across officework and development benchmarks. Anthropic also says the model includes new cyber safety guardrails, reflecting growing attention to model security as capabilities expand. The newsletter also highlights a recent safety paper coauthored by AI leaders from Anthropic, OpenAI, and academia warning about the possibility of selfimproving systems that could accelerate AI progress quickly. It notes that Anthropic’s internal data suggests AI is taking on a larger share of the company’s research and development work. Separately, the piece mentions a separate deal in which AMD agreed to acquire World Labs in an $8.2 billion allstock transaction, bringing FeiFei Li into AMD as chief scientist.