VoicesSimon Willison 8-Post Day - Sonnet 4.6 Review and Rodney v0.4.0simonwillison.net·high signalXBlueskyLinkedInCopy linkWillison published 8 posts covering Sonnet 4.6 SVG benchmarks, Rodney v0.4.0 browser automation, Chartroom CLI, and Qwen 3.5 analysis.SourceSource pagesimonwillison.net↳ Follow the threadShared entity / Stack layerSimon Willison maps ChatGPT Work and finds a code sandbox with unrestricted internet accessSimon Willison's WeblogShared entitySimon Willison Published a Standalone Reference Site for ChatGPT Work's Tool and Skill SurfaceSimon Willison / Hacker NewsStack layerGraham Dumpleton ships wrapture, and draws a line between AI-produced and AI-designed codesimonwillison.netStack layerSimon Willison Counts 223 Tools and 44 Skills Inside ChatGPT Work, and Says OpenAI Documented None of ThemSimon Willison / Hacker NewsStack layer / Threat patternEight MCPHub CVEs published in one 90-second window, topped by a 9.9 that gives any logged-in non-admin remote code executionNVDStack layer / Threat patternOpenClaw 2.0 Ships as an Open-Source Agent Platform That Installs on Your Existing Subscription Instead of Selling OneOpenClawStack layer / ContrastQwen3.8-Flash-Next Matches a 397B Predecessor on 8 of 14 Benchmarks Using a Ninth of the Training FLOPsarXiv 2608.30320Stack layer / Follow-up threadA head-to-head swap of Qwen3.8-Flash-Next-NVFP4 for the dense 27B finds a new failure mode: it says done and outputs nothingr/LocalLLaMA