ABOUT 8 HOURS AGO • 5 MIN READ

☕️ Everything Revealed @ Google I/O 2025 & The Debut of Claude 4

profile

AI Tangle

AI Tangle provides timely, relevant AI news and tools tailored to help business leaders stay ahead of the curve. Our concise, actionable updates ensure you’re equipped to make informed decisions in a rapidly evolving AI landscape. As part of The AIE Network, AI Tangle connects you with additional resources such as AI Marketing Advantage and The Artificially Intelligent Enterprise for complete AI-driven business transformation.

Google's I/O 2025 conference snatches the spotlight soon after Microsoft's Build 2025 with a stupendous amount of releases - Veo 3, Imagen 4, Project Mariner, Jules, and more.

Other key highlights include:

  • Anthropic releases Claude 4, a family of AI models that it touts as the "best coding models in the world"
  • OpenAI buys out ex-iPhone designer Jony Ive's AI devices startup for $6.5 billion
  • Mistral's recently released Devstral blows other small-size AI coding models out of the water

Join us at AI Tangle as we untangle this week's happenings in AI!

Not long after Microsoft's Build 2025 came to a curtain close, Google opened the floodgates to its I/O 2025 conference - and all the rainfall's worth of news, announcements, and releases that came with it. Most notably, Google has beaten OpenAI and Anthropic to have the world's most expensive AI subscription plan with its US-exclusive Gemini Ultra, costing a whopping $250/mo that includes its brand-new Veo 3 AI video generator, new Flow video editing app, as well as what it calls "Deep Think" for Gemini 2.5 Pro. However, that's not the news ends.

Let's elaborate a bit further.

On top of everything mentioned previously, Google also debuted Imagen 4, the next version of its AI image generator, which, in conjunction with Veo 3, will power the aforementioned Flow app. Google's IO also saw the release of two of the company's highly-anticipated AI agents from 2024, Project Mariner and Jules.

At AI Tangle, we have comprised a comprehensive list of everything else announced during the conference to keep this edition concise:

OpenAI rival and fellow Silicon Valley unicorn Anthropic recently debuted Claude 4, its most powerful group of AI models yet, available now on its Pro, Max, Team, and Enterprise plans. The group currently consists of two models, Claude 4 Opus and Claude 4 Sonnet, with the former allegedly able to work autonomously for nearly an entire business day. Much like Anthropic's past models, Claude 4 is catered to developers and thus sports the world's leading scores on SWE-Bench Verified and Terminal, beating out both Google Gemini's latest version and OpenAI's o3 and Codex-1. As part of the same announcement, Anthropic also made Claude Code generally available and added a heap of new integrations.

OpenAI is acquiring ex-iPhone designer Jony Ive's AI devices startup io for $6.5 billion in an all-equity deal, bringing the hardware-focused team in-house while leaving Ive and his "creative collective," LoveFrom, independent. Ive and LoveFrom will take on "deep creative and design responsibilities across OpenAI and io," collaborating closely with OpenAI's research and product divisions. Following on the heels of its recent $3 billion Windsurf purchase, OpenAI's purchase of io marks its largest-ever acquisition as the company looks to expand into consumer hardware and robotics.

On Wednesday, French AI startup Mistral unveiled its next on-device AI model made for software developers - meet Devstral. Building on Codestral and developed together with All Hands AI, Devstral is an open-source AI model available under the Apache 2.0 license as a research preview with as little as 24 billion parameters, meaning it can "run on a single Nvidia RTX 4090 or a Mac with 32GB RAM." Compared to other models of similar size, Mistral says Devstral comfortably beats the likes of OpenAI's GPT-4.1 mini, Google's Gemma 3 27B, and Anthropic's Claude 3.5 Haiku, at least on SWE-Bench Verified.

A recent report by Reuters claims that US semiconductor bigwig Intel is in the market to consider selling off its non-critical networking and edge unit, sources say. This part of Intel's business is responsible for making chips for telecom equipment and has raked in $5.8 billion in revenue in 2024. Alas, Intel's recently-appointed CEO, Lip Bu-Tan, is looking at different ways of helping Intel refocus on its main business, which is part of the new CEO's strategy to get the company back on track. The company has allegedly already engaged with interested buyers but has not started any formal sales process yet.

In an effort to make building e-commerce shops even easier, Shopify has launched its "AI Store Builder," a generative AI feature that lets merchants create complete online stores by simply typing out descriptive keywords. The tool produces three full layouts, including images and text, automating what was previously a manual, laborious design process. Unlike Shopify's past attempts at AI, this is its first fully integrated solution for store setup. Shopify Vice President Vanessa Lee said the goal was to simplify the process using open-ended inputs rather than drag-and-drop fields.

LM Arena, the well-known crowdsourced AI benchmarking platform, has closed a $100 million seed round that values it at $600 million, Bloomberg reports. Backed by a16z, UC Investments, Lightspeed, Felicis, and Kleiner Perkins, the 2023-founded project allows the AI community to evaluate models from all sorts of research labs and industry bigwigs, be it OpenAI, Google, Anthropic, or others. Previously grant-funded via Kaggle and Together AI, LM Arena has faced criticism in recent times for allegedly gaming the leaderboard, which it "has vehemently denied."

Rask - Rask brings language localization at scale to your fingertips - engage with new audiences in over 130 languages with live dubbing and translation of audio and video content, from podcasts to online lectures.

Concierge - Concierge connects to your favorite apps to allow you to talk and ask questions using natural language - ask questions how you feel comfortable and receive comprehensive, well-written answers.

Lindy - Lindy helps businesses build AI agents in minutes to automate workflows such as email management, customer service calls, sales outreach, and more.

Supernormal - Spend less time writing, polishing, and sharing notes and more time on the work that matters with the help of Supernormal's AI-powered meeting transcriptions.

Why Nvidia's CEO Thinks US Chip Controls Are a Total "Failure" (4-min read)

As US chip export controls run rampant, Nvidia CEO Jensen Huang argues that China's homegrown AI push has only accelerated since then, and it has now even begun eating into Nvidia's former 95% China market share.

🔮 The Artificially Intelligent Enterprise - Troubleshoot Anything with ChatGPT

💡 AI CIO - LLM-Based Code Validation for DevOps

🎯 The AI Marketing Advantage - Google vs. OpenAI: The Fight for Search

📻 AI Confidential Podcast - How AI is Reshaping Enterprise Technology With James Kaplan

Your AI Sherpa,

Mark R. Hinkle

Publisher, The Artificially Intelligent Enterprise (TheAIE) Network

Connect with me on LinkedIn

Follow me on X

AI Tangle

AI Tangle provides timely, relevant AI news and tools tailored to help business leaders stay ahead of the curve. Our concise, actionable updates ensure you’re equipped to make informed decisions in a rapidly evolving AI landscape. As part of The AIE Network, AI Tangle connects you with additional resources such as AI Marketing Advantage and The Artificially Intelligent Enterprise for complete AI-driven business transformation.