What is today’s beat?

RELEASES
🧨 Mistral 3 beats GPT-4o
🧨 Deepseek IQ Test Results

NEWSROOM
🗞️ Authors Guild Launches 'Human Authored' Certification for Books
🗞️ Google's 'Ask for Me' AI Calls Businesses on Your Behalf
🗞️ Intel Cancels Falcon Shores AI Chip Amid Strategic Shift

BUILDER BYTES
⭐️ NVIDIA Releases DeepSeek-R1 on NIM Microservice
⭐️ LangChain - how to evaluate document extraction
⭐️ Browser-Use: Bridging AI Agents and Web Browsers

COMMUNITY
🤩 AI Revolution: Why This Is The Best Time To Start A Startup
🤩 Reality versus Speculation 
🤩 NVIDIA's New AI: 150x Faster Virtual Worlds!

FOLLOW US ON SOCIAL FOR MORE

Your FREE newsletter
 share or subscribe 
to show support

🧨 MISTRAL AI 🧨

Mistral 3 beats GPT-4o

Mistral AI, the only European model provider, has introduced Mistral Small 3, a model optimized for low latency and released under the Apache 2.0 license (open source). This model is good achieving over 81% accuracy on the MMLU benchmark and processing 150 tokens per second.

• Efficiency and Performance
  Matches the performance of larger models like Llama 3.3 70B and Qwen 32B while operating more than 3x faster on equivalent hardware.

• Local Deployment
  Quantized models can run on RTX 4090 or a MacBook with 32GB RAM.

• Specs
  24B paramaters, operates at 150 tokens per second, and optimized for reduced VRAM consumption while maintaining high inference speed.

Mistral Small hits all the benchmark criteria to be a top contender, 100% open source with models on Huggingface for download.

⭐️ RESULTS ⭐️

Deepseek IQ Test Results

It is official (not really) that Deepseek has the IQ of an average human being at about 100. Do you notice the confusing results from OpenAI o1 pro?
That’s all for now.

🗞️ NEWSROOM 🗞️ 

What’s hot in tech right now?

Authors Guild Launches 'Human Authored' Certification for Books
Introduces a certification allowing authors to verify their books as human-written, addressing concerns over AI-generated content. The initiative aims to enhance transparency and celebrate the unique human touch in literature.

Google's 'Ask for Me' AI Calls Businesses on Your Behalf
Introduces an AI feature that contacts local businesses to gather information on services and pricing for users. Currently limited to nail salons and auto shops, it utilizes Duplex technology to facilitate these interactions.

Intel Cancels Falcon Shores AI Chip Amid Strategic Shift
Announces the cancellation of its Falcon Shores AI chip to streamline its product roadmap and focus on more immediate AI data center solutions. This move reflects a strategic pivot in Intel's approach to the competitive AI processor market.

Palona AI for Personalized Customer Engagement
Ex Meta and Google engineers teamed up to create Palona, an agentic AI product designed to enhance customer interactions by reflecting each brand's unique personality and voice.

⭐️ BUILDER BYTES ⭐️

What’s hot for builders right now?

NVIDIA Releases DeepSeek-R1 on NIM Microservice
Available as a microservice, it delivers up to 3,872 tokens per second on NVIDIA HGX H200 systems.

LangChain - how to evaluate document extraction
Demonstrates methods to assess document extraction models, focusing on latency, cost, and accuracy - utilizing LangChain in pipelines.

Browser-Use: Bridging AI Agents and Web Browsers
A Github repo by Microsoft that provides a Python library enabling AI agents to interact with web browsers for tasks like navigation and data extraction. Supports integration with language models and offers tools for browser automation.

🤩 COMMUNITY 🤩

What’s the latest beat?

THANK YOU

Looking to promote your company, product, service, or event?