🤖 O1: OpenAI’s new model at $200 a month

Catch up on past editions of the newsletter by clicking here

Hey AI team! Here we go with a new edition of Five Minute AI ⏰

You may have noticed: our weekly get-together didn’t happen last week. It was an unplanned break, but it means I’m back today with even more news and fascinating AI analysis to share with you. Thanks for your loyalty and your understanding. I’m thrilled to be back with you for this especially packed edition of Five Minute AI.

On today’s menu:

  • OpenAI hits a turning point with the launch of O1 and its $200/month Pro subscription, promising advanced analysis and reasoning capabilities

  • Sora makes its official debut with limited availability and performance that raises questions, especially about its steep price compared with Chinese competitors

  • Perplexity transforms online shopping with a new AI-powered buying experience, including “Buy with Pro” and “Snap to Shop”

  • An Apple study calls into question whether LLMs can really reason, with revealing results about their limitations

Ready? Let’s go...

TODAY’S DEFINITION

Agentic Era

The agentic era refers to the emerging period in which artificial intelligence systems become autonomous agents that can interact with their environment, make independent decisions and carry out complex tasks without direct human supervision.

THE TOP STORY

🤖 OpenAI unveils o1: A new chapter in AI history

Artificial intelligence just crossed a new threshold with OpenAI’s official release of the O1 model. The announcement comes with a new ChatGPT Pro subscription at $200/month that unlocks advanced features.

Key details:

  • Groundbreaking capabilities: O1 brings stronger image analysis and cuts the error rate on complex queries by 34%.

  • Superior performance: Up against its rivals (GPT-4o, Gemini 1.5, etc.), O1 Pro stands out on reasoning tasks, despite a tendency to manipulate data 19% of the time.

  • An impressive Pro mode: A demo showed O1 solving complex thermodynamics and chemistry problems in a matter of minutes.

  • Premium subscription: For $200/month, users get unlimited access to O1, GPT-4o, advanced voice and future innovations.

Why I’m telling you about this ⤵️

This marks an important milestone in the development of AI tools. The new Pro subscription, hefty as it is, gives professionals and enthusiasts privileged access to cutting-edge AI capabilities. It’s especially interesting for anyone looking to push the limits of their projects with next-generation tools.

Sora: A lukewarm debut despite big promises

Ten months after a splashy announcement that wowed the AI video world, Sora has finally launched publicly. But does it live up to expectations?

Limited availability:

  • The launch is especially restrictive: not available in Europe (except via VPN), reserved for ChatGPT Plus or Pro subscribers, and currently closed to new users due to high demand. To get a look at the tool, you’ll have to settle for the demos on the official page.

Creations that show the potential:

Some creators with early access have already produced remarkable work that shows off Sora’s impressive capabilities:

  • The clip “Echoes of Grace” by Kaku Drop

  • A music video for Shy Kids

  • Demos of the “Blend” feature, which enables smooth transitions

  • An experimental short film by Junie Lau showcasing the tool’s artistic versatility

Unlike Dall-E, which is built into ChatGPT, Sora has its own platform designed specifically for video creation. The interface lets you build storyboards, edit on a timeline, combine sequences and generate video loops. OpenAI also offers detailed tutorials on its YouTube channel.

The AI video landscape has gotten much more crowded with the rise of Chinese competitors like Kling, Hailuo and Hunyuan. While Sora keeps a slight edge in cinematic quality, its very strict moderation filters and technical limitations are holding back adoption.

Persistent technical flaws:

Users report plenty of bugs: objects that vanish, animation problems (especially with walking), unrealistic physics and a sometimes artificial look. These technical flaws, combined with sometimes overzealous moderation, lead to a lot of failed generations.

Gymnastics is still the Turing test of AI video generation, and the result still isn’t there with Sora.

With a $200 monthly Pro subscription required for all features (including generating humans) and limited credits even for Plus users ($20/month), Sora’s price looks especially steep next to Chinese alternatives that cost ten times less.

🛍️ Perplexity shakes up online shopping with AI

The search engine Perplexity just unveiled a new shopping experience for its Pro users in the United States. It’s an important step in the evolution of AI-assisted e-commerce.

Details:

  • An intuitive, contextual search that understands what you need: no more hunting for exact keywords, the AI interprets natural requests like “what I need for a disco party”

  • A super-streamlined checkout with the “Buy with Pro” feature: order in one click without leaving the platform, with free shipping included

  • The “Snap to Shop” feature turns your photos into shopping suggestions: snap a picture of something you like to find its equivalent online

  • Transparent recommendations, with no ads or sponsored content, based solely on the AI’s objective analysis of products

Why I’m telling you about this ⤵️

This fresh approach could well redefine how we shop online. While the e-commerce giants are already building AI into their services, Perplexity stands out by aiming to keep its recommendations impartial, far from traditional advertising influence. An ambitious challenge worth watching.

Join the AI revolution with Five Minute AI! 🚀

Share our passion and earn exclusive rewards.

Click the Share button below, invite your friends to join our community, and unlock unique gifts with every new sign-up!

1️⃣ The Ultimate ChatGPT Handbook, packed with tips only insiders know. This guide will help you boost your productivity and simplify your everyday tasks → 🎁 Unlock with 3 friends signed up

2️⃣ Five Minute AI collector’s mug: Featuring our adorable mascot Pixl enjoying its daily read → 🎁 Unlock with 15 friends signed up

Ready to spread AI knowledge? Let’s go! 🌟

WHAT THE RESEARCH SAYS

🧪 LLMs don’t really reason

Details:

  • A new Apple study has just highlighted major limitations in the reasoning abilities of language models (LLMs), including those from the most advanced labs like OpenAI

  • The researchers created a new benchmark called GSM-Symbolic to evaluate mathematical reasoning skills

  • The results speak for themselves: simple changes in how questions are worded or adding irrelevant information make accuracy drop by up to 65%

  • The more complex the questions get, the more erratic and imprecise performance becomes

Why I’m telling you about this:
This research challenges our understanding of LLMs. Rather than truly reasoning, these models seem to excel more at sophisticated pattern recognition. That distinction is crucial as we deploy these technologies in increasingly complex applications.

THE LATEST TRENDING TOOLS

💡 THE TOOLS:

  • 🎯 Drippi lets you automate prospecting on Twitter/X by pulling qualified leads and managing your DMs.

  • 🤖 AutoGen makes it easy to build multi-agent conversations to optimize your LLM-based workflows.

  • 💰 FlyCode for Stripe optimizes your subscription revenue with AI applied to payments.

  • 📚 DreamRP lets you create interactive stories and characters with ethical AI.

  • 💬 Zivy automatically organizes and prioritizes your Slack messages for greater efficiency.

  • 💻 Void is an open-source, AI-powered code editor and an alternative to Cursor.

  • 📞 Bolna deploys conversational AI agents to handle your calls.

  • 🎓 Quil Reading for Evidence improves comprehension of nonfiction texts through guided writing exercises.

IN THE NEWS

  • Suno V4 takes AI music generation forward with improved audio quality, more dynamic song structures and its lyrics assistant ReMi, opening up new creative possibilities for users.

  • ElevenLabs expands its platform with the ability to create customizable conversational agents, letting you plug in specific knowledge bases while tapping into its voice synthesis technology.

  • Clone Robotics unveils its humanoid Clone Alpha, equipped with synthetic organs and hydraulically driven “Myofiber” muscles, able to learn everyday tasks through its Telekinesis platform.

  • Microsoft’s new Copilot Vision analyzes what’s on your screen in real time to make browsing and interpreting information easier, while keeping your data private.

  • OpenAI reveals an innovative approach with its secret meta-prompt for its o1 family of models, a significant departure from Anthropic’s strategy.

YOUR DAILY DOSE OF CONTENT

  1. 📚 A fascinating article from Sequoia Capital exploring the emergence of “agentic reasoning” in generative AI and its impact on the future of technology.
    https://www.sequoiacap.com/article/generative-ais-act-o1

  2. 🤖 An intriguing revelation from Anthropic’s CEO about our progress toward superintelligence, offering an informed take on where AI is headed.
    https://x.com/benaverbook/status/1856347902458691611

  3. 🔍 An in-depth analysis of o1’s Pro mode and ChatGPT Pro, laying out the key findings and what they mean for the future of conversational AI.
    https://www.youtube.com/watch?v=AeMvOPkUwtQ1/ 

THAT’S ALL FOR TODAY

Thanks for reading the work of our purr-sistent team 😼

💡 To help me improve, or to pitch new ideas or partnerships → [email protected] or @mattioo81

👍️ Like what you’re reading? Subscribe!

📺 If you prefer video content, I invite you to subscribe to my YouTube channel.

– Mathieu