VoicesClaude Sonnet 4.6 Launch - Opus Performance at Sonnet PricingCNBC·high signalAnthropic launched Sonnet 4.6 with Opus 4.5-comparable performance at /M tokens. SWE-bench 79.6%, now default free-tier model.SourceSource pageCNBC↳ Follow the threadShared entity / Stack layerCross-model 'prompt laundering' and GuardFall shell-injection flaws hit major coding assistantsThreat-Modeling.comStack layer / ContrastChinese Open Models Take Up to 46% of US Developer Token Use as Costs SurgeCNBCStack layer / Threat patternArmadin details full sandbox escape in Anthropic's Claude Cowork; Anthropic disputes, silently patchesSiliconANGLEStack layerMeta Ships Muse Image, Superintelligence Labs' First Image ModelCNBCPolicy dependency / Stack layerAnthropic Restores Global Access to Fable 5 and Mythos 5, Launches Claude for Government BetaAnthropicStack layerMicrosoft commits $2.5B and 6,000 experts to a 'Frontier' deployment unit as AWS backs a $1B forward-deployed teamCNBCStack layer / Threat patternMeta Stands Up a Cloud Business to Sell AI Compute — CoreWeave Stock Drops 14%The Motley FoolPolicy dependency / Stack layerAnthropic opens Claude for Government betaAnthropic