Welcome, AI RESEARCHERS.

The open-source AI model that’s making waves.

OpenAI isn’t the only player in town anymore. DeepSeek’s open-source R1 model is taking on OpenAI’s o1 by delivering top-notch reasoning at a fraction of the cost. Let’s dive in!

In today’s AI Researches:

• DeepSeek R1 Challenges OpenAI’s o1 Dominance

• Some major AI developments

• OpenAI readies ‘o3-mini’ model launch

• Create Seamless Patterns and Apply Them to Product Mockups

• AI Researches & AI pointers

• Trending Ai tools of the week

• More Ai news and developments

Read time: 6 minutes

DEEPSEEK

🐋 DeepSeek R1 Challenges OpenAI’s o1 Dominance

AI Researches: Chinese AI lab DeepSeek has introduced DeepSeek-R1, an open-source reasoning model that rivals OpenAI's o1 in accuracy on major benchmarks—at just a fraction of the cost for developers.

The Details:

• DeepSeek-R1 is built to excel in complex reasoning tasks, especially in fields like physics, math, and science, with an approach similar to OpenAI's o1.

• Massive Yet Accessible: The model boasts 671B parameters, but smaller "distilled" versions, starting at 1.5B parameters, can even run locally on basic hardware like laptops.

• Performance Powerhouse: It outperforms o1 on critical benchmarks, such as AIME, MATH-500, and SWE-bench Verified, showcasing its advanced capabilities.

• Affordable for All: Developers can use R1 for as little as $0.14 per million tokens, compared to OpenAI's hefty $7.50 per million tokens. The model is freely available under an MIT license for commercial use.

Why this Matters: DeepSeek-R1 isn’t just about performance—it’s a philosophical shift. In a world where AI innovation is increasingly locked behind closed doors, DeepSeek’s open-source approach rekindles the original mission of transparency and accessibility. This could spark a new wave of democratized AI, shaking up the balance of power in the AI industry.

An entirely new way to present ideas

Gamma’s AI creates beautiful presentations, websites, and more. No design or coding skills required. Try it free today.

Major Quick Developments

Sam Altman plans to showcase OpenAI's “PhD level” AI agents before U.S. lawmakers in a closed-door session, reflecting the company's cautious optimism about these high-level problem-solving systems and their potential impact on economic and societal initiatives.

UBTech's humanoid robots, starting with the Walker S1, are being deployed in Foxconn's iPhone assembly lines in China to make iPhones, marking a shift toward automation in manufacturing.

Google parent company Alphabet is now creating real world matrix through DeepMind, led by Tim Brooks, former head of OpenAI’s Sora project. These simulations aim to train robots for complex tasks and enhance gaming and VR experiences

Omi, an $89 brain-reading AI device, delivers always-on listening, real-time summaries, and intent detection features, with over 250 open-source apps and AI models from OpenAI and Meta integrated into its ecosystem.

OPENAI

🚀 OpenAI readies ‘o3-mini’ model launch

AI Researches: OpenAI CEO Sam Altman confirmed the upcoming release of the 'o3-mini' reasoning model, a streamlined AI designed to deliver faster responses. The launch will include both API access and ChatGPT integration.

The Details:

• The 'o3-mini' model follows the well-received o1 model, which made strides in science, coding, and mathematics, but trades off some capability for speed.

• Unlike previous staggered rollouts, both API and ChatGPT users will get access to o3-mini at the same time.

• Altman teased a focus on higher-tier models like o3 and o3-pro, hinting at expanded features for users subscribed to the $200/month Pro plan.

• Altman described o3-mini as “worse than o1 pro at most things, but FAST,” suggesting its utility for tasks where speed trumps depth.

AI TUTORIAL

✅ How to Create Seamless Patterns and Apply Them to Product Mockups Using AI

AI Researches: Recraft AI empowers you to transform basic designs into seamless patterns and easily apply them to product mockups, ensuring a professional, on-brand look.

Step-by-Step Guide:

• Step 1
  Sign up for a free Recraft AI account to access 50 daily credits.
  Click “Create New” → “Image,” then search “Seamless Digital v3” for the specialized pattern model.

• Step 2
  Enter a detailed prompt (e.g., “vibrant floral pattern”) while keeping a 1:1 aspect ratio.
  Hit “Recraft” to generate multiple pattern options and select your favorite.

• Step 3
  Duplicate your chosen pattern four times, align them seamlessly, and click “Merge Objects.”
  This produces a larger design while preserving perfect repeats at the edges.

• Step 4
  Select “Create New” → “Mockup,” choose your product (e.g., T-shirt), and maintain a square aspect ratio.
  Click “Recraft” for a blank product canvas ready for pattern placement.

• Step 5
  Drag the merged pattern onto the product mockup—smart placement aligns it automatically.
  Adjust scale and positioning, then preview on various product colors for maximum versatility.

Hire Ava, the AI SDR & Get Meetings on Autopilot

Ava automates your entire outbound demand generation process, including:

• Intent-Driven Lead Discovery

• High Quality Emails with Waterfall Personalization

• Follow-Up Management

Free up your sales team to focus on high-value interactions and closing deals, while Ava handles the time-consuming tasks.

Book a demo to see how Ava can 10x your outbound.

AI Research Roundup

According to new data from the Pew Research Center, ChatGPT usage for schoolwork among U.S. teens has doubled to 26% since 2023, with over 79% now aware of the platform.

A Nigerian study funded by the World Bank revealed that AI tutoring, paired with teacher guidance, helped students achieve two years of learning in six weeks, boosting English, digital skills, and exam results, with a notable impact on girls and potential for greater gains with longer programs.

The University of Bristol unveiled Isambard-AI, a $276M supercomputer that will soon become the UK’s most powerful, using AI to simulate how molecules interact for rapid drug and vaccine development while repurposing waste energy to heat nearby areas.

AI Pointers

Sam Altman says he now thinks a fast AI takeoff is more likely than he did a couple of years ago, happening within a small number of years rather than a decade

Yoshua Bengio says data centers will become a military target once they can run AGI and nations that do not possess that level of AI may be tempted to use nuclear weapons to defend against that level of weaponry

• 📷 Frames by Runway - AI image model offering enhanced realism and precise stylistic control.

• ⚙️ Devin 1.2 - Cognition's upgraded AI coding assistant with better context understanding, a browser-based workspace, and Slack audio integration.

• 🖼️ Krea AI 3D - Transform any image into a 3D object, ready for use in different scenes.

• 🎙️ MiniMax T2A-01 HD - Text-to-audio model enabling voice cloning with just 10 seconds of audio and support for 17+ languages.

• 🖌️ Humva - Design free personalized avatars with thousands of templates for social media, product launches, and more.

• 🎞️ Ray2 by Luma Labs - AI video model delivering enhanced realism and smooth motion.

• 📈 Microsoft 365 Copilot Chat - New free tier with AI agent integrations for productivity tools.

• 🤝 Persana AI - AI-powered GTM and voice agents to boost sales and marketing efficiency.

Runway has launched its Frames AI model, offering photorealistic image generation with advanced stylistic control for creative workflows.

OpenAI CEO Sam Altman clarified on X that the company has not developed AGI and has no plans to release it next month, stating that “Twitter hype is out of control again.”

ByteDance launched Trae, an AI-powered macOS coding tool offering automated project creation and interactive development support in both Chinese and English.

Liquid AI introduced LFM-7B, a language model based on LFM architecture, optimized for advanced chat capabilities in languages like Arabic and Japanese.

Codeium released Windsurf Wave 2, an updated AI development platform featuring web search functionality, automated learning patterns, improved code execution, and enhanced enterprise features.

Character AI unveiled two interactive games, marking a strategic move toward integrating entertainment with AI-driven chatbots.

Cognition Labs rolled out version 1.2 of its Devin AI coding assistant, featuring enhanced context understanding, a browser-based workspace, new enterprise account options, and Slack audio integration.

Krea AI introduced a new image-to-3D object feature in its creative suite, enabling users to transform generated images into editable 3D objects for seamless integration within the company’s real-time tool.

Minimax unveiled T2A-01-HD, a text-to-audio model capable of voice cloning from 10-second samples, supporting over 17 languages, and offering emotional synthesis.

Ndea, a new AGI lab founded by François Chollet and Mike Knoop, focuses on combining deep learning with program synthesis to create efficient, human-level AI, positioning itself as an alternative to large-scale deep learning methods.

OpenAI has introduced ChatGPT Tasks, enabling users to schedule reminders and recurring actions, expanding ChatGPT's functionality into agentic capabilities.

Share Your Opinion