Hi {{firstName|Futurist}},
Last week, Anthropic rolled out a new ad. One that pokes fun at OpenAI, while confidently stating they would never show ads within Claude themselves. A bold claim. Especially coming from a company that chose the biggest ad stage of the year. Because that’s the twist: this ad is scheduled to air during Sunday’s Super Bowl. And by now, that timing feels familiar. The Super Bowl has become Anthropic’s moment to show up with something new. This year was no exception. A fresh model release, right on cue. Of course, OpenAI didn’t sit still. They rarely do. And as expected, they found a way to lean into the moment and grab a share of the attention too.
So grab your favorite snack, settle in, and let's dip into what's cooking. No time to read? Listen to this episode of Digital Dips on Spotify and stay updated while you’re on the move. The link to the podcast is only available to subscribers. If you haven’t subscribe already, I recommend to do so.
🍟 Crispy bites
Fresh tech nuggets. Short, sharp, snackable.
Claude Opus 4.6 raises the bar for knowledge workers
TL;DR
Claude Opus 4.6 is out, a leap forward in AI assisting humans with complex tasks in finance and consulting. By focusing on real-world applications like financial modeling and presentations, Opus 4.6 delivers improvements over 23% stronger than its predecessor. Opus 4.6 is their best model to date. Alongside, Claude Code introduces agent teams, enabling multiple AI agents to work in parallel, boosting productivity. The launch coincides with strategic moves, including Anthropic's Super Bowl ads and a firm stance against ad-based monetization, positioning them against rivals like OpenAI.
Why this matters
Opus 4.6 surpasses Sonnet 4.5 by over 23% on finance tasks
Introduces Claude Code's agent teams, enhancing task parallelization
Embeds AI capabilities directly within tools like Excel and PowerPoint
Anthropic launches high-impact marketing campaigns this week
Stocks of SaaS applications declined post-upgrade announcement
OpenAI challenged as Anthropic underscores ad-free vision
My Taste
Although it feels like Claude Opus 4.5 was only just released, this launch feels practical. It shows AI is moving into everyday tools, turning AI-driven output into the default. Last year, we saw AI reshape coding through “vibe coding”. Now, the shift is happening again. This time, it’s knowledge work. How we think, write, analyse, and decide. This release pushes every knowledge worker a step further. A striking example comes from Rakuten. They found that Claude Opus 4.6 was able to manage a 50-person organization across six repositories, and knew exactly when to escalate to a human. On another note, Anthropic's SuperBowl ads and ad-free emphasis are savvy moves, putting them firmly on the competitive map against OpenAI. And all of this happened in the same week both companies shipped new models. We’re watching AI learn to do more things at once. Faster. Closer to how humans actually work.
OpenAI ships a Codex app alongside GPT-5.3-Codex
TL;DR
The newly launched GPT-5.3-Codex achieves state-of-the-art coding performance on benchmarks like SWE-Bench Pro and Terminal-Bench. It excels in managing professional knowledge work and is significantly faster than its predecessor, GPT-5.2. As the first model trained to identify software vulnerabilities, it also benefits from enhanced cybersecurity capabilities. These advances suggest a more efficient and secure computing environment. At the same time, OpenAI released a standalone application called Codex, allowing developers to spin up multiple parallel agents.
Why this matters
Achieves 57% on SWE-Bench Pro and 76% on TerminalBench 2.0
Matches GPT-5.2 in professional knowledge work on GDPval
Reduces token use by over 50% from GPT-5.2, accelerating interactions
Shows 64% performance on OSWorld-Verified across multiple OS platforms
First model directly trained for identifying software vulnerabilities
My Taste
What stands out here is the model's dual focus on improving both coding and cybersecurity. The fact that early versions of GPT-5.3-Codex were used to accelerate its own development is a vivid illustration of AI aiding its own evolution. This self-reinforcing development cycle shows where the future of AI is heading. With the release of the Codex app, managing these advanced capabilities should become more intuitive. Expect a paradigm shift, not just in speed and performance, but in how we collaborate with AI in real time, as steering conversations while they unfold becomes the norm.
OpenAI's Frontier enters the enterprise AI battle
TL;DR
OpenAI has launched Frontier, a service allowing businesses to build and manage AI agents within their infrastructure. The move targets expanding AI's practical uses, like automating repetitive tasks, directly challenging rivals like Anthropic. Frontier aims to streamline AI integration, making it easier for companies to adopt without overhauling existing systems. Expect tough competition in the enterprise AI space, with OpenAI eager to reclaim territory.
Why this matters
Frontier lets companies create AI agents compatible with existing tools.
AI agents can handle tasks like software debugging, reducing human effort.
CEO Sam Altman emphasizes business growth as a key focus for OpenAI.
Rival Anthropic's Claude models present direct competition.
Frontier aims to centralize AI management, minimizing disparate tool use.
My Take
OpenAI's introduction of Frontier signals a strategic shift to regain enterprise AI ground, especially against Anthropic's increasing focus on business applications. I see this as OpenAI's attempt to shed the consumer-only perception by offering a platform that doesn't just multiply AI usage but democratizes it. By simplifying AI adoption for non-techies, OpenAI's approach is a bet on self-service and internal scale. This could very well redefine the boundary between technical and operational capabilities in businesses. I see this as a build-up to what you open first when you flip open your work laptop. Instead of email, you open your AI workspace. It’s what I wrote last week: you don’t need scale. You are the scale.
Everyone a director with Kling 3.0
TL;DR
Kling 3.0 is launching its latest multimodal creative engine, promising unprecedented ease in video, audio, and image production. The model ensures high video realism in short 15-second clips, supports multiple languages and accents in audio, and delivers cinematic visuals in 4K images. It's a leap forward for creatives seeking realistic and customizable outputs, erasing the line between AI-generated content and reality.
Why this matters
Kling 3.0 offers precise control in video production, enhancing content quality.
Supports audio in multiple characters, languages, and accents.
Produces 4K images with new cinematic features.
Allows flexible video production with realistic 15-second clips
Provides tools for realistic, short-form video creation, changing creative workflows.
My Taste
It's pretty wild to think about, but now almost anyone can produce high-quality videos. Professional-grade media production is now accessible to those well outside traditional creative circles. It feels almost surreal to say that AI-generated commercials can now look as real as any shot on a high-end camera. This lines up with a broader trend we've seen: technology empowering more people to take on roles traditionally filled by specialists. We’re entering an era where storytelling is democratized; you don't rely on big budgets or teams. This could change how brands communicate, putting traditional production houses on notice. Let's see which industries embrace this new standard first.
🧀 Cheesy pick
A cheesy selection of three tools and one tasty rabbit hole.
🍱 Leftovers
A roundup of updates that are too cheesy to ignore.
Google’s LiteRT's universal framework turbocharges on-device AI with 1.4x faster GPU performance and unified NPU acceleration.
Google’s Agentic Vision enhances Gemini 3 Flash’s prowess in analyzing complex images, perfect for reading fine details.
Google’s Gemini Enterprise's new Marketing Agents transform real-time data into revenue-driving campaigns.
OpenAI’s ChatGPT unleashes over 60 new apps, with more rolling out soon.
Gamma lets you generate AI animations for presentations with Leonardo 2 and Veo 3 styles on Business and Ultra plans.
ModelScope’s LingBot-World debuts an open-source world simulator for next-gen video generation with real-time, high-fidelity graphics.
Mureka V8 turns AI-generated music into release-ready tracks with human-like logic and expressive vocals.
Invideo AI Motion Graphics collaborates with Anthropic AI to let small teams create pro-level motion design from a single prompt.
Freepik Multiple Model Generation lets you test up to 4 models at once to quickly find your ideal AI solution.
Freepik Spaces introduces List Nodes for instant, large-scale content generation and localization.
Anthropic’s Cowork empowers Claude with plugins to tailor skills and commands for your team.
Anthropic Claude AI drops an official connector for Supabase, tackling security issues effortlessly.
xAI’s Grok Imagine 1.0 unlocks 10-second videos in 720p with vastly improved audio quality.
ElevenLabs’ Eleven v3 exits alpha, boasting improved stability and 68% fewer errors on technical details.
Zai GLM-OCR dominates document analysis with just 0.9B parameters, acing formula and table recognition.
n8n Chat Node v2.5 introduces human-in-the-loop actions for more dynamic agent workflows.
Alibaba Qwen3-Coder-Next debuts with 800K tasks for coding agents and local dev in a sleek 80B-param package.
AceMusic’s ACE-Step-v1.5(2B) tops music generation with local magic, full songs under 10s, and free commercial use.
Lovart Skills empowers your design dreams with automated, professional-grade workflows.
Mistral’s Voxtral Transcribe 2 revolutionizes speech-to-text with MistralAI's sub-200ms real-time transcription.
Skywork 2.0 ditches credit limits for unlimited creativity on Windows desktop.
Attio transforms your CRM interactions with smart searches and automatic follow-ups by just ‘Ask Attio’.
Perplexity's Model Council lets Max subscribers compare outputs from three frontier models for more accurate web answers.
Perplexity's Deep Research supercharges your insights with top-tier accuracy and reliability, now accessible to Max and soon Pro users.
Topazlabs’ Starlight Fast 2 upgrades video to pristine 4K with double speed and unlimited free access.
Cursor showcases long-running coding agents with over 1,000 commits per hour in a special research preview.
World Labs showcases their Real-Time Frame Model which can explore in real-time, interactive frame generation from images.
v0 lowers barriers with an all-in-one platform for coding and deploying software instantly.
How’d this digital dip taste?
This was it. Our forty-nine digital dip together. Forward this to someone who thinks AI is for coders, not doers.
The tools aren’t the hard part anymore. Knowing how to use them is. If you’re not sure where to start, or how to go from dabbling to doing, I can help. Whether you're a solo builder, a growing team, or a whole organization, I’ll help you turn today’s AI options into real leverage. Let’s make AI your advantage. Just reply to this email.
Looking forward to what tomorrow brings! ▽
-Wesley
