🏃‍♂️Mistral Joins The Reasoning Race

Today we will discuss:

• Sponsored: MyReport—Turn any topic into a citation-ready report in minutes

• Mistral enters reasoning AI with Magistral launch—The models are built for logic-heavy tasks but don’t outperform top models in early tests

• Don’t miss this week’s Workflow Wednesday—build ethical AI workflows using ChatGPT to draft a real ethics policy, SEMrush to automate content responsibly, and GitHub to build a transparent workflow; now available at only $20/month! 

• o3-pro Is OpenAI’s new top model—Outperforms rivals in reasoning, adds tool use, now live for Pro users

• In other AI news—OpenAI delays its open model launch to after June, Zuckerberg is personally assembling a new “Superintelligence” team at Meta to pursue AGI, OpenAI reportedly signs a major cloud deal with Google despite their rivalry, and China plans to turn Shenzhen into an AI and aviation hub to counter US sanctions

• Saved the best for last—10 must-try AI tools

Write Reports in Minutes, Not Hours

Tired of spending your life formatting citations and googling pie chart data?
Meet MyReport - the AI tool that turns your topic into a fully researched, citation-rich document in just a few clicks.

✅ Describe your topic
✅ Review the draft
✅ Get a polished report with sources, visuals, and formatting

Whether you’re submitting an assignment, pitching to investors, or creating content that actually converts, MyReport does the heavy lifting for you.

Plagiarism-free. APA/IEEE ready. Internet-sourced and human-level smart.

Try it free - first draft on us.

Mistral steps into reasoning AI with new Magistral model family

Key Points 

• Mistral released two reasoning models with the larger one (Magistral medium) available in preview on Le Chat and its API.

• Benchmark tests show Magistral medium underperforming compared to top models like Claude Opus 4 and Gemini 2.5 Pro.

📯News - French AI startup Mistral is entering the reasoning model space with a new family of models called Magistral. These models are designed to tackle problems step by step, improving accuracy in areas like math, physics, and logic.

It’s a move that puts Mistral in the same category as OpenAI’s o3 and Google’s Gemini 2.5 Pro, though not quite at the same level just yet.

🤓Here's the lowdown - There are two versions of Magistral. Magistral small has 24 billion parameters and is available to download on Hugging Face under an Apache 2.0 license. Magistral medium is more powerful and is currently in preview through Mistral’s Le Chat chatbot its API, and select cloud partners.

While Mistral says the models are well suited for enterprise tasks like structured calculations, decision trees, and logic workflows, its own benchmarks show Magistral Medium lagging behind. On reasoning and programming tests like GPQA, AIME, and LiveCodeBench, it underperforms compared to Claude Opus 4 and Gemini 2.5 Pro.

🤔So what makes it stand out? What Mistral is betting on instead is speed and accessibility. The company says Magistral delivers answers up to 10 times faster on Le Chat and supports a wide range of languages, including Italian, Arabic, Russian, and Simplified Chinese.

Read the full story here.

 Workflow Wednesday #23 just dropped: For people using AI who’d like to do it ethically 

This week’s theme: AI Ethics

ChatGPT – Build a Legit Ethics Policy
 SEMRush – Automate Content Responsibly
 GitHub – A Responsible AI Workflow

The future won’t wait. Build workflows that last.
 Join the movement 

o3-pro replaces o1-pro in ChatGPT with major upgrades

Key Points 

• o3-pro is now live for ChatGPT Pro and team users replacing o1-pro and expanding to API access

• o3-pro beats top models from Google and Anthropic on internal math and science benchmarks according to OpenAI.

👨‍💻News - OpenAI has officially released o3-pro, the latest version of its reasoning-focused AI model. It’s now available to ChatGPT Pro and team users with access for enterprise and Edu customers coming next week. The model is also live in the API starting today.

🤩What's new & noteworthy? Unlike typical AI models that rely heavily on pattern recognition, o3-pro is built to work through problems step by step. This makes it more reliable in fields like math, physics, and coding. It replaces the older o1-pro model in ChatGPT.

According to OpenAI, expert reviewers consistently preferred o3-pro over the base o3 model across all tested areas. It performed especially well in science, education, business, and writing help. It also ranked higher for clarity, comprehensiveness, and instruction-following.

The model can use tools like Python, web browsing, file analysis, and visual reasoning. It also supports memory-based personalization. The trade-off is that it responds a bit slower than o1-pro.

Internally, o3-pro scored higher than Google’s Gemini 2.5 Pro on AIME 2024 and outperformed Claude 4 Opus on GPQA Diamond, which tests PhD-level science. However, it still can’t generate images and doesn't support Canvas and temporary chats are currently disabled while OpenAI fixes a technical issue.

Read the full story here.