Anthropic Open Sources Skills Standard as OpenAI and Google Unveil New AI Models
Anthropic transforms its Claude Skills into an open standard for portable AI workflows. Meanwhile, OpenAI launches GPT-5.2-Codex for advanced coding and Google introduces Gemini 3 Flash for rapid, cost-effective reasoning.
Anthropic makes its Skills feature an open standard so that repeatable AI workflows are portable across different tools and platforms. The company also launches a directory of pre-built skills from major tech companies like Notion, Canva, Figma, and Atlassian. New administrative features allow organizations to provision these skills directly from settings, making it easier to create and edit automated workflows.
OpenAI releases GPT-5.2-Codex, a version of its latest model optimized specifically for coding agent tasks. This update brings better context compaction for long-horizon work, stronger performance on large code refactors and migrations, and improved cybersecurity capabilities. The model is currently available to paid ChatGPT users across Codex surfaces, with an API rollout planned after further safety improvements, alongside a new invite-only program for vetted cybersecurity professionals.
Google introduces Gemini 3 Flash, a new frontier model designed to deliver fast reasoning at a lower token cost. This model excels in iterative development and high-frequency workflows, outperforming both Gemini 2.5 models and Gemini 3 Pro in coding capabilities on the SWE-bench Verified benchmark. Its strong performance in reasoning and tool usage makes it a highly efficient option for developers.