DeepSeek Crashes the AI Party

Another day, another new reasoning model…

DeepSeek is Making The Giants Nervous…

DeepSeek has unveiled its R1 Lite model in a limited preview, and it’s already making waves in the AI community.

The model, accessible only via chat for now, is impressively rivaling—and in some cases surpassing—OpenAI’s o1-preview model. It showcases “chain-of-thought” reasoning (aka, mental gymnastics), transparently documenting its problem-solving steps to explain what it’s doing and why.

While some reasoning paths may seem odd, the model excels at tackling tricky questions and even outperforms GPT-4o and Claude in certain tasks.

It can even answer the infamous strawberry question:

Early tests show R1 Lite surpassing OpenAI’s o1 in popular benchmarks like MATH and AIME, achieving scores 12–18% higher:

So What?

DeepSeek has also released data demonstrating that its model becomes more accurate when allocated additional time or “thought tokens” for problem-solving. The results show consistent improvements in benchmark performance, including AIME, as the model processes inputs with greater depth.

In other news, yes, we've got another AI model to add to our increasingly crowded model family tree. But hey, I'm not complaining. Everyone gets 50 free messages per day on the chat interface, try it out yourself by smashing the button below!

Scrape any data from any website, with AI

FetchFox is the easiest web AI scraper. Visit any website, type your scrape prompt and click "Run". The powerful FetchFox backend can scrape a 1 page or 1,000 pages without skipping a beat. You can scrape popular sites like Reddit, X, LinkedIn, Google results and more.

The Great Showdown - ChatGPT Search vs. Perplexity

As a die-hard Knicks fan, I figured what better way to test these AI search features than with some basketball stats?

I put ChatGPT Search and Perplexity through their paces with a simple prompt:

Give me the score to the last Knicks game and tell me how they are doing this season.

The results? Well you’ll have to see for yourself…

Looks like ChatGPT’s search won out here! Perplexity actually ended up referencing the “last” game of the previous season, but I was looking for the score of “last night’s” game.

While Perplexity usually nails these queries, it seems to have dropped the ball this time. Maybe my prompt was too vague, or maybe ChatGPT's new search feature is bringing its A-game? 🤔

Put it to the test 👇️

Two Approaches to Designing Your AI Team with CrewAI

Before we dive in, let's get everyone on the same page. AI agents are autonomous software programs designed to perceive their environment, make decisions, and take actions to achieve specific goals.

Think of them as specialized AI workers with their own roles, tools, and decision-making capabilities. Unlike basic AI models that simply respond to prompts, agents can:

• Use tools and APIs

• Execute multi-step tasks

• Make autonomous decisions

Now, imagine coordinating a team of these AI workers. That's where CrewAI comes in. The package offers hierarchical and sequential processes, two distinct approaches to agent task management tailored to different project needs, here is a breakdown:

Using Sequential saves resources for predictable tasks (like processing forms), but struggles when you need real-time decisions. Hierarchical offers flexibility for complex tasks but adds unnecessary compute and complexity if a simple sequence would do the job faster and cheaper.

Stay updated with our favorite highlights, dive in for a full flavor of the coverage!

Mistral unveiled new AI models and the Le Chat chatbot platform can now search the web — with citations in line.

Google released LearnLM, an experimental task-specific model that has been trained to align with learning science principles when following system instructions for teaching and learning use cases.

Tessl raises $125M at $500M+ valuation to build AI that writes and maintains code.

Meta hires Salesforce’s CEO of AI, Clara Shih, to lead new business AI group.

Mustafa Suleyman, CEO of Microsoft AI, says AI can have “infinite memory” by 2025.

Elon Musk’s xAI is raising up to $6 billion to purchase 100,000 NVIDIA chips for Memphis data center.

OpenAI just shipped a new update that makes GPT-4o more creative.

Trending Tools

🍐 Pear AI - An open source AI code editor.

💥 Particle News - An AI-powered news app that organizes and summarizes news articles, providing users with multiple perspectives and customizable summaries for efficient news consumption.

📹️ Google Vids - An AI-driven video presentation tool that transforms text prompts into video presentations, automatically incorporating stock footage, generating scripts, and providing AI voiceovers to facilitate content creation.

This warmed my heart, ChatGPT being ethically useful for homework…

If you have any suggestions or specific feedback, simply reply to this email or fill out this form. Additionally, if you found this insightful, don't hesitate to engage with us on our socials and forward this over to your friends!

You can find our past newsletter editions here!

This newsletter was brought to you by Digestible AI. Was this forwarded to you? Subscribe here to get your AI fix delivered right to your inbox. 👋