World's First 'Hybrid Reasoning' AI Model?

It’s Tuesday. Welcome to the first newsletter of AI Execs. Moving forward, the format may change, but for now…we’ll keep it simple. Think of the current edition as my personal notes on AI, personally delivered to you.

Here’s the quick summary:

• Anthropic announced Claude 3.7 Sonnet

  • “Our most intelligent model to date and the first hybrid reasoning model on the market.”

  • “We’re also introducing a command line tool for agentic coding, Claude Code. Claude Code is available as a limited research preview, and enables developers to delegate substantial engineering tasks to Claude directly from their terminal.”

  • “Users can toggle ‘extended thinking mode’ on or off, directing the model to think more deeply about trickier questions.”

    • AI “thinking” has been discussed for years. It reminds me of this recent paper from a Stanford team about test-time scaling, which covers interesting performance results from simply having LLMs “wait”.

    • Here’s another paper on test-time scaling from a Johns Hopkins teams.

• Perplexity is launching a new browser called Comet

  • They’re calling it “a browser for agentic search”

  • No release date, can join the waitlist

The numbers:

• OpenAI’s weekly active users surpass 400 million (RT)

• Apple will spend more than $500 billion in the U.S. over the next four years (Apple)

• Microsoft reiterates plan to invest $80 billion in AI (CNBC)

• Alibaba announced plans to invest at least $53 billion over the next three years to advance its cloud computing and AI infrastructure (Alizila)

• Anthropic is finalizing a massive $3.5 billion funding round (WSJ)

What else?

• Apple preparing Google Gemini integration with Apple Intelligence (9to5Mac)

• Salesforce and Google bring Gemini to Agentforce (Salesforce)

• Spotify opens up support for ElevenLabs audiobook content (Spotify)

• YouTube for Android TV gains a ‘dubbed’ tag for AI translated audio (9to5Google)

• AI models have, within hours, created more efficient wireless chips through deep learning (Live Science)

• xAI announced Grok 3 Beta (X)

• Epilepsy AI tool detects brain lesions doctors miss (BBC)

• Figure’s humanoid robot takes voice orders to help around the house (TC)

Thought starters:

• Satya Nadella (Microsoft CEO) recently discussed his vision for AI-enabled knowledge work in a conversation with Dwarkesh Patel (Onur Solmaz blog)

• Refresher of Dario’s (Anthropic’s CEO) philosophy of AI, Machines of Loving Grace, from back in Oct 2024.

• The AI Firms of The Future Will Be Copyable (Forbes)

Thanks for reading,

Eddie

P.S. If this was valuable, please share it with your friends. It all starts with knowledge and curiosity - let’s keep exploring together.