• Horizon AI
  • Posts
  • Groq's AI Chip Powers the World's Fastest AI ⚡

Groq's AI Chip Powers the World's Fastest AI ⚡

+ 'Mistral Next' as good as GPT-4?

Sponsored by

Welcome to another edition of Horizon AI,

Groq’s new AI hardware (unrelated to Musk’s ‘Grok’) has been making waves on X for its unprecedented speed. Could this new tech position itself as a compelling alternative to Nvidia’s GPUs?

Let's get started!

Read Time: 3.5 min

Here's what's new today in the Horizon AI

  • Groq Unveils Lightning-Fast Processor for Near Real-Time Text Generation ⚡

  • Mistral's Next LLM Rivals GPT-4? Try it now and find out!

  • AI Research: AI Girlfriends Failing to Protect User Privacy, Study Finds

  • AI Tutorial: Get GPT-4 Vision in your phone camera

  • AI Tools to check out

  • The Latest in AI and Tech 💡

AI News

GROQ

Groq Unveils Lightning-Fast Processor for Near Real-Time Text Generation

Source: Groq

AI startup Groq has developed a new specialized hardware 'LPU' (Language Processor Units) that significantly accelerates the performance of large language models in text generation tasks.

Details:

  • Unlike conventional GPUs, which boast hundreds of cores tailored for graphic rendering, LPUs are specifically designed to ensure deterministic performance for AI computations.

  • LPUs can generate up to 500 tokens per second, far surpassing the 30-50 tokens per second achieved by other popular AI models like GPT-3.5 running on traditional hardware.

  • X users have compared its speed with that of GPT 3.5, yielding impressive results.

While Groq's LPUs are currently limited to inference of pre-trained models, this specialized hardware could provide a compelling alternative to NVIDIA's high-end GPUs that presently dominate the AI market.

Test Groq's remarkable speed using Mixtral and Llama here

MISTRAL

Mistral's Next LLM Rivals GPT-4? Try it now and find out!

Mistral, the French AI startup, is preparing to launch its next major language model, which could potentially rival GPT-4. For now, a prototype is available for testing.

Details:

  • ‘Mistral Next’ is expected to be their largest and most advanced model yet, aimed at closing the gap with GPT-4.

  • A Mistral developer confirmed on Discord that the startup will officially launch the model soon.

  • Early testers have reported very positive impressions, suggesting it could perform at the level of GPT-4.

  • You can try out Mistral Next for yourself right now by accessing the "Direct Chat" interface on Chatbot Arena.

Despite its early stage, Mistral Next is generating considerable excitement as a formidable competitor to OpenAI's GPT-4. It will be interesting to see if this French AI startup's latest model lives up to the high expectations.

Your SOC 2 Compliance Checklist from Vanta

Are you building a business? Achieving SOC 2 compliance can help you win bigger deals, enter new markets and deepen trust with your customers — but it can also cost you real time and money.

Vanta automates up to 90% of the work for SOC 2 (along with other in-demand frameworks), getting you audit-ready in weeks instead of months. Save up to 400 hours and 85% of associated costs.

Download the free checklist to learn more about the SOC 2 compliance process and the road ahead. 

AI RESEARCH

AI Girlfriends Failing to Protect User Privacy, Study Finds

New research from the Mozilla Foundation has revealed concerning privacy practices among popular AI chatbots that offer users a romantic relationship. According to the study, none of the 11 chatbots reviewed, including well-known apps like Replika and Eva AI, met the organization's safety requirements, putting them "on par with the worst categories of products" ever evaluated for privacy by Mozilla.

The researchers identified several "red flags" regarding these chatbots' handling of personal data. Notably, they found that the apps fail to encrypt user information to meet minimum security standards, potentially exposing sensitive details to unauthorized access.

The organization advises users to avoid disclosing sensitive details and to request the deletion of their data once they stop using the app. Furthermore, users should be cautious about granting these chatbots continuous access to their location data, photos, videos, and device cameras.

AI Tutorial

Get GPT-4 Vision in your phone camera

By following these steps, you'll be able to use GPT-4 to obtain descriptions of things around you simply by tapping the back of your iPhone twice.

  1. Get an OpenAI key.

  1. Add this iPhone Shortcut to your phone

  2. Enter your OpenAI key inside the Shortcut where it says “Text”.

  3. In order to access the Shortcut by double-tapping on the back of your phone:

Go to Settings > Accessibility > Touch, and tap Back Tap. Next, tap Double/Triple Tap. Lastly, select your Shortcut name.

  1. Give it a try

Open your camera, point it at what you want to know about, and tap the back of your phone two or three times

Source: @dr_cintas on X

AI Tools to check out

‍🎓 Repeto: Unlock smarter and faster learning with AI.

 Prompty: Provide your prompt for analysis and optimization using proven techniques.

🎨 Playform: A no-code AI art generation platform with a variety of tools for creative people.

📃 Text Humanizer: Enhance your content with our AI to achieve the quality of human-authored writing.

✍🏻 TaxGPT: Simplify your tax season with personalized answers, maximized refunds, guaranteed accuracy

The latest in AI and Tech

Neuralink's first human patient implanted with a brain-computer interface has fully recovered and is now able to control a computer mouse using only their thoughts, according to Neuralink founder Elon Musk.

Microsoft is set to significantly expand its AI and cloud infrastructure in Spain with a $2.1 billion investment over the next two years. The company claims it is not just about building data centers but committing to helping develop the country’s “security, and development and digital transformation of its government, businesses, and people.”

Singapore's AI ambitions just got a major boost with a $740 million investment plan announced by Deputy Prime Minister Lawrence Wong. Over the next five years, the country aims to secure access to advanced AI chips, work with leading companies to establish AI centers of excellence, and further enhance its AI capabilities.

ElevenLabs is expanding its portfolio with a new text-to-sound model capable of generating sound effects based on textual descriptions. While the model is not yet publicly available, ElevenLabs has demonstrated its capabilities through a teaser featuring videos enhanced with AI-generated sounds. The company has established a signup page and is inviting potential users to join an early access waitlist for the upcoming model.

Nvidia has replaced Tesla as the most actively traded stock on Wall Street, highlighting the increasing significance of AI investments in the market. This follows the chip maker's recent rise to become the third-most valuable U.S. company.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

We'd love to hear your thoughts on today's email!

Your feedback helps us improve our content

Login or Subscribe to participate in polls.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻