The last edition was only two weeks ago. Yet in those two weeks, a lot has changed. Not only in the world itself, but especially in the world of AI. Last Friday, Anthropic said it did not want its models used by the US Department of War. A day later, OpenAI stepped in. The reaction was fast. And loud. Criticism flooded in. Some users even cancelled their ChatGPT subscriptions. Meanwhile, Anthropic became the most used AI model in the US, also among organizations. In a recent interview, Anthropic’s CEO Dario Amodei said: “We do not see hitting a wall. This year will have a radical acceleration that surprises everyone.” Exponentials tend to catch people off guard. And according to Amodei, we are standing at the edge of something big.

Let’s take a look at what’s on the menu: cheaper brains, faster models, and images in 4K for free. With enterprise agents knocking on SaaS doors and one system running 19 models at once, the benchmarks are shifting by the week.

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.

Fresh tech nuggets. Short, sharp, snackable.

Claude Sonnet 4.6 marks a notable upgrade in AI model capabilities, now making advanced operations more accessible and cost-effective. This update excels in coding, computer use, and long-context reasoning with a 1M token context window in beta. At just $3 per million tokens, it challenges the need for more expensive models. The model shows remarkable progress, approaching the performance of high-end systems for tasks that previously needed specialized tools.

Read it yourself

Take a look at the demo video. Anthropic slipped the real update into a simple demo. Picture this: a small business owner types "make delivery free for orders over $25" into the chat, and the model handles the rest. No code, developer, or complex instructions needed. That's the whole SMB automation market flipped in seconds. At a fraction of the price per action, cost becomes a non-factor for AI adoption. For every small business owner, suddenly the question isn’t “can I afford AI?” but “why am I still doing this myself?”. And with a 1M token context window, the model can hold the whole shop in its head at once.

Google has rolled out two powerful models: Gemini 3.1 Pro and Gemini 3.1 Flash-Lite. Gemini 3.1 Pro brings advanced reasoning to complex tasks, doubling the intelligence of its predecessor at no extra cost. Meanwhile, Flash-Lite offers speed and cost-efficiency, outperforming previous models. Together, these advances reshape how AI integrates into daily applications, setting new expectations for both intelligence and affordability.

Read it yourself

The benchmark table says more than the keynote. The quiet number? $0.96 per ARC-AGI-2 task. From 31.1% to 77.1% in three months. Same $2 per million input tokens. Smarter, same price. Crazy. I was thinking about what I read online: “Look closely at the scoreboard. Google leads ARC-AGI-2 at 77.1% with Gemini 3.1 Pro. Anthropic tops SWE-Bench Verified at 80.8% with Opus 4.6. OpenAI takes Terminal-Bench at 77.3% with GPT-5.3-Codex. And Opus grabs Humanity’s Last Exam with tools at 53.1%. No one sweeps the board. The ‘best model’ badge now rotates weekly. Each lab owns a different column of the same sheet.” I could not said it better myself. Because, the true revelation here is the speed at which Google can translate research breakthroughs into production models. Moving reasoning advancements from Deep Think to Gemini 3.1 Pro in just seven days is more telling than any individual metric. Cost remains static while intelligence scales, marking a shift where raw AI capability becomes a baseline expectation. Google's infrastructure gives them an advantage, but the clock is ticking as competitors tighten the gap. In this dynamic landscape, whoever integrates intelligence into everyday life most seamlessly is poised to lead.

Google's new Nano Banana 2 image model is out, highlighting a blend of speed and precision in image generation. It offers real-world accuracy, enhanced linguistic capabilities, and visuals with vibrant colors and textures. This launch is poised to redefine digital design amidst rising content demands, delivering powerful tools at no additional cost. With features supporting up to 4K graphics and precise text rendering, the upgrade is a significant step forward in accessible digital artistry.

Read it yourself

Nano Banana 2 is a stunning reminder of how quickly technology can level the playing field. With zero cost and high capability, it's the ultimate toolkit for anyone from hobbyists to professionals. Designers may find their traditional roles challenged as the power to create high-quality visuals becomes widely accessible. The future here isn't about hiring experts; it's about empowering individuals to become creators. We're witnessing an era where tools once gated by skill and cost become lightning-fast and easy for anyone to use.

Anthropic has rolled out its enterprise agents program, leveraging AI to bring pre-built agents into everyday workplace tasks. The offering includes plugins for finance, legal, and HR, aiming to fulfill the promise of agentic AI in corporations. These agents promise to simplify integration within companies and could significantly disrupt SaaS solutions currently in use. The move is part of a broader push to make agentic AI a staple in enterprises, following slower-than-expected adoption last year.

Read it yourself

Anthropic is cranking up the heat in the AI sector, shaking up stock markets left and right. They've managed to trigger several selloffs, with this release being the latest catalyst. This isn't just a product update; it's a shift in thinking about AI deployment in businesses. Companies adopting these agents aren't just upgrading their current AI tools, they're redefining their processes. Anthropic's rapid-fire announcements are reminding SaaS companies that the future waits for no one. They're forcing a hard choice: adapt or become irrelevant. As history shows, businesses that choose to partner usually fare better. Those resisting change? They're often left counting losses.

Perplexity Computer claims to combine every current AI capability into one cohesive system. It manages to research, design, code, deploy, and monitor projects seamlessly. In the race towards an all-in-one system solution, its ability to orchestrate 19 models in parallel stands out, potentially redefining how tasks are handled. The implications for removing middlemen in project management and coordination are significant, shifting focus to individuals with a knack for creativity and context.

Read it yourself

Perplexity Computer epitomizes the ultimate AI utility. It's like they took OpenAI's vision and made it universally accessible, using every available model to its full potential. As AI evolves to handle end-to-end processes, those with vision and understanding will call the shots and lead projects solo. You might remember the ‘hype’ around OpenClaw from a few weeks back? This is OpenClaw for everyday users.

A cheesy selection of three tools and one tasty rabbit hole.

A roundup of updates that are too cheesy to ignore.

This was it. Our fifty-one digital dip together. Forward this to someone who’s waiting for AI to slow down.

If you’re thinking about rolling out Anthropic’s enterprise agents, or applying any of the shifts in this edition inside your company, I can help you design it, set it up, and make it actually work in practice. Just reply to this email.

Looking forward to what tomorrow brings! ▽

-Wesley