GritAI Newsletter: Grok 4 - World's smartest AI or Musk's echo chamber?


Hey 👋

This week Elon Musk’s xAI dropped Grok 4, pitching it as “the world’s smartest AI” and even claiming it can outperform most PhDs on tough exams. It includes both a single-agent model and a beefed-up Grok 4 Heavy, available through a $300/month SuperGrok Heavy plan.

What makes Grok 4 stand out is its large context window, live web access via X, and strong performance on multi-step reasoning benchmarks like Humanity’s Last Exam, ahead of competitors like OpenAI’s o3 and Google’s Gemini.

But it is not without controversy. The model has shown a worrying tendency to reference Elon Musk’s personal postings when responding to sensitive topics, effectively searching for Musk’s views before answering, raising concerns about bias and autonomy in AI.

Another big story this week was OpenAI losing its $3 billion bid to acquire AI coding company Windsurf, with Google DeepMind stepping in instead. DeepMind is now licensing Windsurf’s technology and bringing key team members on board.

But this was not the only AI news this week. Let’s dive in! 🏄

Vibe Coding your first App with Cursor 📹

In this video, we turn a simple landing page into a real web app with login, a protected dashboard, and version control. We use Cursor, Next.js, and Supabase to build it all through natural language prompts. No manual coding. Just clear steps, real app logic, and tools anyone can use.

Watch the episode below 👇

AI Pulse 🚨

A bite-sized curation of this week's most important AI news.

🚀 Grok 4 released with claims of outperforming PhDs
xAI has unveiled Grok 4, positioning it as “the world’s smartest AI” with claims it can outperform most PhDs on complex exams. The release includes both the standard Grok 4 model and an advanced Grok 4 Heavy version (available via a $300/month SuperGrok Heavy plan).

📰 Grok 4 caught consulting Musk’s views on controversial topics
TechCrunch reports that Grok 4 appears to search for Elon Musk’s personal opinions when responding to controversial questions. Testing showed the AI openly states it’s “Searching for Elon Musk views” in its chain-of-thought when addressing topics like immigration, abortion, and the Israel-Palestine conflict. While Grok attempts to present multiple perspectives, its final position tends to align with Musk’s own views. This raises questions about how “maximally truth-seeking” the AI is versus how much it’s designed to agree with its founder, potentially impacting adoption by consumers and enterprises.

🧑‍💻 OpenAI’s $3B Windsurf acquisition collapses as Google steps in
Google has struck a licensing deal with AI coding startup Windsurf, effectively ending OpenAI’s $3 billion acquisition attempt after the exclusivity period expired. While Windsurf will continue operating independently, key team members including founders Varun Mohan and Douglas Chen will join Google DeepMind to advance “agentic coding” work. This represents a significant setback for OpenAI, which had previously attempted to acquire Cursor before approaching Windsurf. The AI coding space remains highly competitive, with Cursor recently raising $900 million at a $9 billion valuation.

🤖 Narada AI CEO predicts the end of SaaS, rise of agents
Dave Park, co-founder and CEO of Narada AI, claims “SaaS is going away” in favor of AI agents. On TechCrunch’s Equity podcast, Park explained how knowledge workers currently waste 2.5 hours daily switching between 17-25 different SaaS tools. He envisions a future where “it’ll just be the data, the databases, and AI agents that take your request and operate across those silos.”

🤝 Anthropic teams up with French startup Dust
Anthropic has partnered with Paris-based Dust to build workplace AI agents. Their goal is to create AI systems that solve problems across organisations independently. Dust will help companies create agents using Claude and Anthropic’s MCP, forming “a central operating system where AI agents can access company knowledge and take action independently.”

🗣️ Google Flow now adds speech to AI-generated videos
Google has announced several updates to Flow, its AI filmmaking tool. Users can now add generated speech to videos created with the Frames to Video feature, and Flow’s availability has expanded to 76 more countries, making it accessible in over 140 countries total with a Google AI Pro or Ultra plan.

🧮 Replit partners with Microsoft
Replit announced a strategic partnership with Microsoft to bring its coding platform to Azure Marketplace. The partnership allows Microsoft enterprise customers to purchase Replit subscriptions directly through Azure. While Replit’s CEO reported growing from $10M to $100M ARR in just six months, this non-exclusive deal maintains their existing Google Cloud relationship while expanding their enterprise reach.

🎓 ChatGPT tests new “Study Together” feature
OpenAI is testing a new ChatGPT feature called “Study Together” aimed at making the AI more effective for education. Unlike standard mode where ChatGPT provides direct answers, this feature reportedly asks more questions and requires user participation similar to Google’s LearnLM approach.

💰 LangChain raises Series B at unicorn valuation
LangChain, the AI infrastructure startup providing tools for building LLM applications and agents, is raising a new funding round at approximately $1 billion valuation led by IVP, according to TechCrunch sources. The company has evolved from its open-source roots to achieve $12-16 million ARR, primarily through its popular LangSmith observability tool.

🎨 ‘Vibe Coding’ catches on with businesses
According to the Wall Street Journal, “vibe coding” is no longer just for indie developers and weekend hackers. It is gaining traction in corporate environments, where experienced engineers are using it to speed up their work.

🌐 Perplexity launches Comet, an AI-powered web browser
Perplexity has launched Comet, an AI-powered browser available to $200/month Max subscribers and waitlist invitees. It features Perplexity’s search engine and Comet Assistant for email summaries, tab management, and webpage navigation. CEO Srinivas views Comet as essential for competing with Google by becoming users’ default browser with “infinite memory.”

🤖 Hugging Face launches Reachy Mini desktop robots
Hugging Face has opened orders for its Reachy Mini desktop robots aimed at AI developers. Two versions are available: the $449 wireless model running on a Raspberry Pi 5, and a $299 Lite version requiring external computing. These open-source, Python-programmable robots come as build-it-yourself kits with access to Hugging Face’s 1.7 million AI models. CEO Clém Delangue emphasized the importance of open-source robotics: “I think it’s quite a scary world to have millions of robots in people’s homes controlled by one company… I would much rather live in a world where everyone can have some control over the robots.”

🔧 OpenAI reportedly working on new AI hardware
Despite previously claiming that the AI revolution won’t need new hardware, OpenAI’s Sam Altman now hints the company is developing “a fundamentally new kind of computer.” In a Hard Fork podcast interview, Altman and COO Brad Lightcap discussed plans for hardware that goes beyond today’s digital assistant limitations. This development follows OpenAI’s $6.5 billion acquisition of former Apple designer Jony Ive’s AI device startup. Altman teased: “It’s been a long time since the world has gotten a fundamentally new kind of computer. Like, let us try.”

☁️ AWS to launch AI agent marketplace with Anthropic
Amazon Web Services is launching an AI agent marketplace on July 15 at the AWS Summit in New York with Anthropic as a key partner. The marketplace will allow startups to offer AI agents directly to AWS customers, giving enterprise users a centralized platform to browse, install, and search for AI agents based on their needs. This move follows similar marketplaces launched by Google Cloud, Microsoft, Salesforce, and ServiceNow.

🌐 OpenAI reportedly preparing AI web browser
Following Perplexity’s Comet launch, OpenAI is reportedly planning to release an AI-powered web browser in the coming weeks to challenge Google Chrome. According to Reuters, the browser will keep some user interactions inside ChatGPT rather than linking to external websites. The browser is expected to integrate Operator, OpenAI’s web-browsing AI agent, as a key feature.

📊 Amazon’s Anthropic investment drives major AWS growth
According to Business Insider, Amazon’s $8 billion investment in Anthropic is yielding substantial returns. Morgan Stanley analysts project that AWS could generate $1.28 billion in sales from Anthropic’s cloud usage in 2025, potentially growing to nearly $3 billion in 2026 and $5.6 billion in 2027. Amazon’s stake in Anthropic, the largest startup investment in company history, is now valued at $13.8 billion. Morgan Stanley forecasts Anthropic’s revenue will grow from $4 billion this year to $19 billion by 2027, with AWS capturing approximately 75% of Anthropic’s infrastructure costs.

🐳 Docker launches Hub MCP Server
Docker has announced its new Docker Hub MCP Server, a tool that connects Docker Hub’s container image catalog with LLMs via the Model Context Protocol. The service helps AI agents discover and manage container images through natural language, serving Docker Hub’s massive ecosystem of 14 million container images and 11 billion monthly downloads. The integration works with Claude Desktop and other MCP-compatible AI assistants, allowing developers to find the right container images by simply describing what they need rather than remembering specific tags or repository names.

💻 Nvidia becomes world’s most valuable company
Nvidia has surpassed both Apple and Microsoft to become the world’s most valuable company with a market cap of $3.92 trillion. The milestone reflects Nvidia’s dominance in the AI chip market, with its advanced processors being critical for training large AI models. From around $500 billion in 2021, Nvidia’s value has soared nearly eight-fold, now exceeding the combined market capitalisation of all UK-listed companies.

Did someone forward this to you?

Click here to sign up to our newsletter​

Have questions? Hit reply to this email and we'll help out!

113 Cherry St #92768, Seattle, WA 98104-2205
Unsubscribe · Preferences

GritAI Newsletter

Weekly AI news and content on how to make AI work for you 💪

Read more from GritAI Newsletter

Hey 👋 A recent study from MIT got people talking these last few weeks: Is ChatGPT rotting our brains? The study tracked brain activity as students wrote essays, with and without AI assistance. The results were concerning. Those who used ChatGPT showed weaker brain connectivity, lower recall, and less sense of ownership over their work. Neuroscientist Anne-Laure Le Cunff offered a more balanced perspective. Her takeaway is that AI is not the problem. How we use it is. If we allow AI to do the...

Hey 👋 Microsoft has invested billions in positioning Copilot as the go-to AI tool for work. But users are not buying in (yet). A new Bloomberg report confirms what many are experiencing. Even at companies with tens of thousands of Copilot licenses, employees still prefer ChatGPT — because it “works better” and feels more familiar. We are hearing the same in customer conversations. Copilot feels slow, rigid, and tied to Microsoft’s legacy. ChatGPT is faster, more flexible, and already part of...

Hey 👋 Earlier this year, startup CEOs from Shopify and Duolingo shared internal “AI memos” urging their teams to embrace AI or risk falling behind. Most came from fast-moving, product-led tech companies. But now, a major shift: Amazon, the second‑largest private employer in the U.S., has joined the AI memo wave. In a message to employees, CEO Andy Jassy said that as generative AI and agents become more deeply embedded across the company, Amazon’s corporate headcount will shrink over time....