Moonshot AI released free download weights for its 2.8-trillion-parameter sparse MoE with a 1M-token context window and ~50B active params per token; API pricing is $3/$15/MTok and self-hosting clears data-residency concerns no closed-API model can.
Why it matters: Teams handling sensitive data now have a credible frontier-class open model to self-host — meaningful for any operator weighing open vs. closed model risk.
Cognizant embeds Claude Code into its Flowsource and Neuro® platforms, making it the default coding agent for Cognizant's Fortune 500 delivery network.
Why it matters: Claude Code's enterprise reach expands through Cognizant's existing client base — accelerating AI engineering adoption for large enterprises that buy through IT services firms.
Dario Amodei draws a clean line: Anthropic supports mandatory safety testing and chip export controls, not open-weight bans, directly addressing policy debates that affect enterprise model procurement decisions.
Why it matters: Useful framing for any operator evaluating open vs. closed model strategy — Anthropic's actual position is less restrictive than widely reported, backing testing requirements over bans.