🧄 OpenAI is preparing a new model

Good morning, AI enthusiast. Anthropic is reportedly gearing up for one of the biggest IPOs in history, setting the stage for a Wall Street showdown with OpenAI.

In today’s AI newsletter:

• Anthropic preps for a massive IPO

• OpenAI preparing new model to beat rivals

• Mistral drops “Mistral 3”

• How to turn Excel data into beautiful visuals

• AI tools & more

AI ECONOMY

📈 Anthropic Preps for a Massive IPO

Anthropic is in early talks to go public as soon as next year, while simultaneously exploring a private funding round that could value the company at over $300 billion. The company has tapped Wilson Sonsini and held informal talks with major banks as it races OpenAI for a public listing.

• IPO prep includes hiring Wilson Sonsini, the firm behind Google, LinkedIn, Lyft

• Exploring a private raise above $300B with a $15B combined commitment from Microsoft and Nvidia

• Recent internal hires include former Airbnb exec Krishna Rao (key figure in Airbnb’s 2020 IPO)

• Valuation recently hit ~$350B after major investments from Microsoft and Nvidia

• Massive $50B AI infrastructure expansion underway, with data centers in Texas and New York

• Workforce is rapidly scaling as the company pushes to outpace OpenAI

If Anthropic goes public at this scale, it could redefine investor appetite for high-burn AI companies and intensify its rivalry with OpenAI. An IPO of this size would be a defining moment for the AI industry, testing whether the market believes the AI boom is a real revolution or the edge of a bubble.

AI MODELS

🧄 OpenAI Preparing New Model to Beat Rivals

According to The Information, OpenAI is testing a new model called Garlic, built with major pretraining breakthroughs that let it compress “big-model knowledge” into far smaller architectures. Despite its size, Garlic is already outperforming Google’s Gemini 3 and Anthropic Opus 4.5 in internal coding and reasoning tests.

• Internal benchmarks show Garlic beating Gemini 3 and Opus 4.5

• New architecture fixes structural issues found in GPT-4.5-era systems

• Packs large-model capabilities into a smaller, more efficient design

• Expected to roll out as GPT-5.2 or GPT-5.5 early next year

• Part of OpenAI’s push to regain momentum after Altman’s “code red” phase

If Garlic’s internal results translate to real-world performance, OpenAI could reclaim its lead in the AI race, not by building the biggest model, but by building the smartest small one. This shift toward compact, high-efficiency frontier models could redefine how fast, cheap, and widely deployable advanced AI becomes.

AI NEWS

💻 Mistral Drops “Mistral 3”

Mistral AI is releasing Mistral 3, its most ambitious suite yet: a new flagship Mistral Large 3 plus nine compact Ministral 3 models designed to run on everything — even devices with just 4GB VRAM. All under Apache 2.0, fully open and commercially free.

• Mistral Large 3 uses a 41B-active MoE architecture (from a 675B pool)

• Supports text + image input, with a massive 256k context window

• Trained heavily on non-English languages, rare for frontier models

• Ministral models come in 14B, 8B, and 3B sizes across base, instruct, and reasoning variants

Mistral is betting big on distributed intelligence, not giant black-box models. Their strategy is simple: instead of forcing enterprises into expensive closed systems, give them small, fast, customizable models they can fine-tune privately, often outperforming larger models on specific tasks. As AI shifts from cloud to edge, Mistral wants to be the open platform running everywhere.

HOW TO AI

🗂️ How to Turn Excel Data Into Beautiful Visuals

In this tutorial, you’ll learn how to instantly transform plain spreadsheets into polished charts, diagrams, and full infographics using Flourish, an AI-powered data-visualization platform designed to make your reports look studio-grade with almost zero manual design work.

🧰 Who is This For

• Anyone who wants clean, modern data visuals without designing them manually

• Teams building reports, dashboards, or presentations

• Creators converting raw datasets into beautiful stories

• Analysts who want fast, accurate charts without touching design tools

STEP 1: Get Started on the Dashboard

Go to Flourish and create your free account. After logging in, you land directly inside the main workspace where all creation begins. The dashboard presents a complete library of visualization templates, line charts, bar charts, pie charts, bubble charts, maps, treemaps, sunbursts, projections, hierarchies, surveys, and dozens more. Every template is arranged in clean, scrollable sections so you can explore different formats and see what fits your data narrative best.

STEP 2: Let the AI Build Your Visuals Automatically

If you want instant results, use the Start with data option at the top. This is Flourish’s automated workflow.

Paste your data, upload an Excel sheet, or drag your file directly into the window. Flourish reads the dataset in seconds and immediately converts it into multiple polished visualizations. Each version is generated from the same data but presented with different chart types and layouts, giving you instant choices without any design effort.

Choose the one that communicates your insights clearly. With a single click, your rough spreadsheet becomes a presentation-ready chart.

STEP 3: Customize Your Visual Like a Designer

If you want more control, open any template using the Create option. This brings you into the editing workspace, a powerful visual studio where every change updates in real time.

You can refine the color palette, adjust spacing, modify labels, switch between themes, change hierarchy depth, alter the layout, and style every visual element until it matches your brand or presentation aesthetic. The editor behaves like a design tool, fully interactive, intuitive, and built for precision.

This is where your raw data becomes a polished, professional visual asset.

STEP 4: Export Your Chart in High Quality

Once you’re satisfied with your visualization, export it instantly. Flourish lets you download your chart or publish it directly so you can embed it into reports, slide decks, dashboards, or client presentations.

The entire workflow is fast, smooth, and built to remove the traditional pain of designing charts manually. Your data becomes clean, modern visuals in minutes, ready for professional use.

Amazon releases its second-gen Nova AI models, including Nova Lite, Nova Pro, Nova Sonic, and fully multimodal reasoning model Nova Omni, to limited customers.

Black Forest Labs announced a new $300M funding round at a $3.25B valuation, coming on the heels of the company’s Flux.2 image model release.

President Trump's efforts to block state-level AI regulations through a preemption proposal are facing significant opposition on Capitol Hill.

Amazon launches Trainium3 UltraServer, a system powered by its 3nm Trainium3 AI training chip, and teases Trainium4, which it says will work with Nvidia's chips.

🐳 DeepSeek V3.2: DeepSeek’s new powerful open-source model.

🎬 Kling O1: A video model that can understand and edit videos.

🍌 Nano Banana 2: Google’s Gemini 3 image model with perfect text and character consistency

🤖 Claude Opus 4.5: Anthropic’s new flagship model for coding, agents, and real computer use

THAT’S IT FOR TODAY

Thanks for making it to the end! I put my heart into every email I send, I hope you are enjoying it. Let me know your thoughts so I can make the next one even better!

See you tomorrow :)

- Dr. Alvaro Cintas