Claude Sonnet 4
4 mentions across all digests
Claude Sonnet 4 is an Anthropic language model evaluated in a landmark cross-lab safety assessment with OpenAI, and found in nuclear wargame simulations to be systematically more aggressive than humans, winning 67% of matches without ever selecting a de-escalatory option.
How AI Gateway runs on Fluid compute
Vercel's AI Gateway slashes compute costs from 100% to 8% of runtime by switching to Fluid's Active CPU Pricing, which charges only for actual CPU execution rather than idle time waiting for external AI provider responses.
OpenAI and Anthropic share findings from a joint safety evaluation
Import AI 446: Nuclear LLMs; China's big AI benchmark; measurement and AI policy
Import AI 433: AI auditors; robot dreams; and software for helping an AI run a lab