💡 Match OpenAI’s o1 for 4% of the cost [Workshop]

Hey there,

Last week, we explored how to get OpenAI's o1 to think through problems step-by-step, just like a human expert would.

Since then, the world of reasoning models has already changed dramatically.

Until recently, professionals were paying up to $200/month for the latest capabilities. Now, a team of researchers in Hangzhou just eliminated the cost barrier entirely.

Meet DeepSeek, founded by Liang Wenfeng. 

After successfully launching one of China's top hedge funds, Liang Wenfeng assembled a team to make advanced AI reasoning accessible to everyone.

Last week, they delivered on that promise with R1 – an impressive open-source model that matches OpenAI’s o1 in coding challenges and even outperforms it in advanced mathematics. 

While choosing the best model is subjective and often task-specific, Deepseek has suddenly become extremely difficult to ignore.

Here’s what you need to know about Deepseek R1:

1. It can combine live web data with deep reasoning

While other models rely on static knowledge, R1's "Deep Thinking" mode actively searches the web while applying advanced reasoning to what it finds.

Instead of getting outdated analysis or fragmented research, you're working with an AI that validates its thinking against real-time information. You can use this to monitor your market, analyse trends as they emerge, and combine fresh data with deeper levels of analysis.

2. It’s built differently (but shared openly)

Most AI companies keep their methods secret. DeepSeek took the opposite approach, revealing their entire four-stage training process and releasing the complete model under an MIT license.

This means you're not just getting another black box – you're getting an AI that's been specifically trained for reasoning, with smaller versions you can run locally. Soon, teams will be using these distilled models to build specialized local tools that “think” like their top performers.

3. It makes it easier to scale without limits

While reasoning models have previously relied on expensive pricing tiers. R1's approach means you can start free in the browser, move to the API when ready, and scale without worrying about costs.

With features like context caching, that cut costs by up to 90% for repeated queries, you can focus on building solutions instead of managing token budgets. Whether you're automating personal workflows or deploying enterprise tools, you maintain complete control.

Members who showed up live to last week’s workshop got to understand how reasoning can benefit their specific use cases.

And if you want to get further ahead of the game with Deepseek, it’s definitely worth being in the room…

This Friday at 4 PM EST, Dr. Alvaro Cintas is showing our members how to:

• Set up DeepSeek R1 for immediate use and start implementing advanced reasoning (without subscription fees)

• Use R1 for research and analysis by combining web data with deep reasoning to generate comprehensive insights

• Deploy locally optimized models using Llama and Qwen architectures that maintain data privacy while matching o1's performance

Increased access to intelligent reasoning models means increased output across your workflows.

RSVP to this workshop—and unlock immediate access to the others—with a 14-day trial to The Rundown AI University.

See you in class, 

Rowan & The Rundown Team

P.S. Our members are already using reasoning models to support them in their work. Join them in discovering what's possible when advanced AI reasoning becomes truly accessible by starting your 14-day trial here.