Grok Code Fast 1

A lightweight model built for agentic coding.

Welcome to AI Wire — your smart shortcut to all things AI, without the jargon.

What we’ll cover today:

🕵️ Anthropic wants your chats for training

💸 Nvidia is printing cash from GPUs

🎓 MathGPT is changing how students cheat

⚡ Grok Code Fast is free (for now)

  • New drop: xAI launched grok-code-fast-1, a lightweight model built for agentic coding.

  • Why it’s cool: It’s trained on real pull requests, tuned for IDE workflows, and runs tool calls (grep, terminal, file edits) fast.

  • Speed boost: Uses caching and new serving tricks so it feels instant, even in loop-heavy coding tasks.

  • Wallet-friendly: $0.20 per million input tokens, $1.50 per million output with free access on GitHub Copilot, Cursor, and others (for now).

  • Anthropic is asking Claude users to either allow their chats to be used for training or opt out by September 28.

  • If users agree, the company will store their data for up to five years.

  • Business customers are excluded from this policy, so only regular consumers are affected.

  • The real reason behind the change is Anthropic’s need for large amounts of real-world data to compete with OpenAI and Google.

If your AI chats could be used for model training, what would you do?

Login or Subscribe to participate in polls.

FROM OUR FOUNDER…

The Smartest Way to Grow Your Newsletter 🚀

Tired of low-quality subs and wasted ad spend? I’ve scaled AI, fintech, SaaS, and multiple niche specific newsletters with real, engaged and high intent subscribers. Would love to do the same for you!

 40%+ open rates: no bots, no junk leads.
 No ad management fees: just pay per engaged subscriber.
 First-party data & proprietary methods ensure quality.
 Risk-free scaling: we’ve built and grown newsletters ourselves.

📩 Want high-intent subscribers that actually engage? Let’s talk.

  • Nvidia reported $46.7 billion in revenue last quarter, which is 56 percent higher than the same time last year.

  • The company’s profit hit $26.4 billion, driven mainly by demand for its AI chips.

  • Its newest Blackwell chips alone generated $27 billion in sales.

  • Even though sales to China are blocked, Nvidia still expects to make $54 billion next quarter.

  • MathGPT is expanding from 30 schools to more than 50 universities this fall.

  • The AI tool never gives students direct answers but instead guides them with questions to reduce cheating.

  • Professors can set rules, use auto-grading, and even require proof of student work.

  • The tool costs $25 per student and is being marketed as a safe and reliable AI for classrooms.

🛠️ TOOLS

🤖 OpenAI Codex – One agent for everywhere you code.
Grok Code Fast 1 – Speedy economical model for agentic coding.
📅 Wanderboat – AI finds best local dining and events.
📓 HyNote – AI note-taking with instant meeting summaries.



📚 RESOURCES

🔍 Ultimate NotebookLM Guide – Google’s AI notebook explained.
💸 Apple Intelligence Setup – Upgrade your iPhone with AI.
⏱️ Grok Code Fast in VS Code – See Grok’s speed inside VS Code.
OpenAI Codex CLI – Run Codex directly from your terminal.

This video explains why the future of programming is not about writing lines of code but about giving clear instructions that AI and people can both understand. Sean Grove shows how writing specs can guide AI, align teams, and even replace code in some cases.

Was this forwarded to you? Sign up here.

AI Wire News. 

Signing off