AI is king. And $TAO is the king of AI. The sooner you accept the 100% fact that AI is the biggest & only thing that matters, the sooner your children will thank you for making them rich!

The best open model outside China did not come from the U.S. It came from Paris. Mistral launched Large 4, nicknamed Le Chonk: one trillion parameters, 49 billion active, factory multimodal. On aggregated benchmarks, it is the strongest open-weights model ever built in the U.S. or Europe—in other words, outside China. Until now, the world’s most capable open models were Chinese. The API is already live. Weights will be released at the end of October. Trained from scratch in Europe on 3,800 Grace Blackwell GPUs within Mistral’s own data centers. Operates under European law. The strike against closed models: in visual grounding, it outperforms closed frontier models. In cybersecurity, on the Artificial Analysis Cyber Index, it ranks in the global top 5 and leads all open models outside China. In a test to reproduce and patch a real-world vulnerability, it scored 82%—the highest of any model. Claude Opus 5.5 and GPT-6 Astra scored near zero, because they refused the task. Defending software often begins by proving a vulnerability exists. Closed models freeze. Open models, in the hands of companies, do not. The battle now is this: the U.S. and Europe in closed models. China in open models. Paris has entered the game with weight.

Claude Code has introduced a new feature called Mods. The core change is that previously, you could only communicate with the underlying model via prompts, CLAUDE.md, or Skills—but the Claude Code application itself (responsible for the UI and executing actions on your computer) was hardcoded by Anthropic. Now, with Mods, you can directly modify the application’s visual interface and underlying execution logic—all without writing any code yourself. Simply describe the functionality you want in conversation, and Claude will write and load it for you. Many users confuse Mods with the existing Hooks, but they are fundamentally different. Hooks are external scripts, like third-party contractors—they run their task and then terminate, with no access to the app’s internal interface or memory. Mods, by contrast, run directly inside the app’s core: they can directly modify or replace Claude Code’s interface, share the same memory space to track token usage or correction counts, and provide interactive buttons to intercept actions. Additionally, Mods support custom shortcuts—for example, typing `/cost` triggers local calculation by the Mod, consuming zero tokens. To obtain and install Mods, simply update Claude Code to the latest version—the feature is enabled by default on both terminal and desktop apps. You can paste a Marketplace link directly into Claude Code to auto-install, or use terminal commands. The community has already compiled over fifty open-source Mods; for example, the official engineer-created “Next Steps” Mod automatically suggests next steps and recommended commands above the input box after a task completes. However, Mods have permissions to read files, execute programs, and access the internet—always verify the behavior of third-party Mods before installing for security. To create your own custom Mod, simply describe your needs to Claude. The Mod becomes active immediately without restarting, but initially runs in temporary mode. Once satisfied, explicitly instruct Claude to save it permanently. All installed Mods are stored in the `.claude/plugins` folder in your system’s root directory. Use the terminal command `claude plugin list` to view them, and remove them via `claude plugin uninstall` or by deleting the corresponding folder directly. This level of底层 openness enables entirely new workflows. For instance, when handling multiple tasks simultaneously, you could create a progress bar Mod that displays remaining time and completion status for all sessions above the input box. Or build a retrospective Mod that, upon detecting user corrections, pops up a dedicated panel to guide review and update Skills and cross-file MDs—preventing redundant instructions. Even community Mods now exist that integrate with LINE for auto-replies or play games during compilation wait times. When even the container layer of a tool can be instantly restructured via natural language, the barrier to customizing software has been completely eliminated.

in chinese AI means “to love” SI means “to die”

My battle against ChatGPT with @TurboFlow_xyz is what pushed me into going live. Thank you to the whole team and to everyone who has followed us since 🙏 A bigger new chapter starts Friday and yes, @grok, your turn is coming soon 👀

The LLMs will likely be stopped by some made up fear. But the more likely truth is that they need an easy exit after promising the world. The real economic activity seems to be not working out. This could have knock-on effects to the other Ai companies.