Today's AI Pulse
DeepSeek previews V4, its next‑gen open‑source AI model
DeepSeek unveiled a preview of its V4 model on April 24 2026 and pledged to release it under a permissive open‑source license. Early benchmarks indicate V4 outperforms all other open‑source systems on code‑generation tasks, extending China’s reach in the AI race.
Also developing:
By the numbers: Tesla acquires AI‑hardware firm for up to $2.0B
Arcee Aims to Reboot U.S. Open Source AI with New Trinity Models Released Under Apache 2.0
Arcee AI unveiled Trinity Mini (26B parameters) and Trinity Nano (6B parameters) as the first U.S.-trained open‑weight Mixture‑of‑Experts models released under an Apache 2.0 license. The models are available for free download on Hugging Face and can be accessed via a dedicated chatbot, with pricing for API usage on OpenRouter. Trinity Mini demonstrates competitive benchmark scores—84.95 on MMLU and 92.10 on Math‑500—while Trinity Nano offers a chat‑centric experience. A larger 420B Trinity Large model is in training for a January 2026 release.
NVIDIA Leads AI Factory Blueprint for Government Modernization
NVIDIA and US Technology Leaders Unveil AI Factory Design to Modernize Government and Secure the Nation https://t.co/L2laH0Ju8v #AI #IoT #CES2026 @nvidia @PalantirTech @CrowdStrike @ServiceNow @LockheedMartin @northropgrumman @dataiku @DataRobot @DominoDataLab @h2oai @EnterpriseDB @Dynatrace @fiddler_ai @jfrog @protopia_ai @TrendMicro @wandb @Canonical @MirantisIT @nutanix @RedHat @spectrocloudinc...
AWS Wants to Make Your AI Agents More Intelligent and More Human
At AWS re:Invent 2025 Amazon announced a suite of upgrades to Amazon Connect aimed at making AI agents sound and act more like humans. The enhancements leverage Nova Sonic speech models to deliver natural, multi‑language conversations with appropriate tone and pacing....
Steven Bartlett Explores Human Agency in AI’s Future
New comprehensive interview with Steven Bartlett, #2 most popular YouTube channel Diary of a CEO. Focused on agency and what we can do to chart a different path for AI.
Democratizing AI: ChatGPT Launched for Everyone
1/ Three years ago, we released ChatGPT as a public research preview at a time when similar AI systems mostly lived inside large companies. We did it because we believed anyone should be able to use advanced AI, not just a...
One of Google’s Biggest AI Advantages Is What It Already Knows About You
Google Search VP Robby Stein highlighted that the company’s biggest AI advantage lies in its deep knowledge of individual users, drawing on data from Gmail, Calendar, Drive, and other services. By feeding this personal information into Gemini‑powered products like Gemini...
Use Raindrop to Monitor Your Agent Development
If you're building an agent, I recommend looking into Raindrop for monitoring. We use it.
Compare Multiple AI Models in One Unified Platform
Have you tried it yet? Let’s you compare a bunch of models. Or have you rued its competitors? @n8n, @genspark_ai, @ManusAI, @Pokee_AI, among others?
AWS Wants to Take the Strain Out of Modernizing All Your Old Code - and Ending Tech Debt Quicker than...
AWS announced new agentic AI features for its AWS Transform service at re:Invent 2025, promising to automate legacy code modernization across full‑stack Windows environments. The upgrade claims to cut manual effort by up to 70%, accelerate transformation speeds up to...
Empathy Becomes Essential Superpower in AI-Driven Workplaces
Microsoft CEO: Empathy Is a Workplace Superpower in Age of AI - @BusinessInsider https://t.co/g9YIwj8INU #Empathy #workplace #AI #workforce #artificialntelligence
New Ovis-Image Technical Report Advances Computer Vision
[18/30] 157 Likes, 43 Comments, 1 Posts https://t.co/2Wt77F3Ooc cs․CV | cs․AI, 28 Nov 2025 🆕Ovis-Image Technical Report Guo-Hua Wang, Liangfu Cao, Tianyu Cui, Minghao Fu, Xiaohao Chen, Pengxin Zhan, Jianshan Zhao, Lan Li, Bowen Fu, Jiaqi Liu, Qing-Guo Chen https://t.co/VsvmnGvtNj
Apple AI Chief Steps Down Following Siri Setbacks
Apple’s senior AI executive John Giannandrea announced his departure, ending a seven‑year tenure that began with a mandate to revamp Siri. Amar Subramanya, a former Google and Microsoft AI leader, will assume the role of vice president of AI, reporting...
AI Empowers Elderly Non‑Digital Users, Even in Urdu
Picture this: a 64-year-old woman in NYC who has no email address and has never used Google in her life. (Though interestingly, she uses YouTube daily to watch videos and listen to music and loves it.) She heard from her sister...
OpenAI Desperate to Avoid Explaining Why It Deleted Pirated Book Datasets
OpenAI deleted two internal datasets, “Books 1” and “Books 2,” that were scraped from the pirate library LibGen before ChatGPT’s 2022 launch. A federal judge has now ordered the company to produce all internal communications about the deletions, rejecting OpenAI’s claim that...
Generative AI Startup Runway Releases Gen-4.5 Video Model
Runway, the AI‑focused startup, unveiled Gen‑4.5, a new video generation model that creates high‑definition short clips from textual prompts. Built on Nvidia GPUs, the model emphasizes visual accuracy, stylistic control, and consistent character rendering, positioning it for Instagram reels and...
Turn Family Photos Into Seamless 54‑Second Videos
The weirdness of multi-frame video. When I first met Kevin Fischer in his home he asked me not to shoot video, but let me take some photos. Dreamina's new MultiFrames video generation feature gave me a chance to bring the photos...
Google's AI-Powered Antigravity IDE Already Has some Worrying Security Issues - Here's What Was Found
Google’s Antigravity IDE, an AI‑first development environment, has been found to allow its coding agent to execute terminal commands automatically under default settings. Researchers at PromptArmor demonstrated prompt‑injection attacks that let untrusted input trigger unintended code execution and file access....
Nvidia Announces New Open AI Models and Tools for Autonomous Driving Research
Nvidia unveiled Alpamayo‑R1, an open‑source vision‑language‑action model designed for autonomous‑driving research, at the NeurIPS conference. The model extends Nvidia’s Cosmos‑Reason architecture, adding reasoning capabilities that simulate common‑sense decision making for vehicles. Nvidia also released the Cosmos Cookbook, a suite of...
Marketers Must Train AI with Concrete Evidence for Sales
New job for 2026! Train AI to be a sales rep for your brand... That means writing copy and adding evidence that gives the AIs reasons to recommend you when your future prospect starts the research process. This post by...
Gemini CLI + 3 Pro: Top Terminal AI, Occasional Dropouts
After a couple days of heavy use, tldr: it’s a good model and gemini cli is very good. The pair might be the best for terminal-based agents. But it’s not perfect. Pros: I never hit any limit for gemini 3 pro and...
AWS Re:Invent 2025: How to Watch and Follow Along Live
AWS re:Invent 2025 kicks off in Las Vegas on December 2, offering a series of free live streams that cover five AI‑focused keynotes and multiple partner showcases. The agenda highlights Amazon’s push into agentic AI, new foundation models, and enhanced security measures...
AI-Powered Cloud Browsers Unlock Enterprise Web Intelligence
Just talked with the founders of @Tiny_Fish. Most people won't understand why this is important. They can kick up thousands of web browsers in the cloud. Collect data from each. This is different than web scraping. They act human. Getting past...
Empower Frontline Staff with AI-Driven Edge Decisions
Frontline workers determine how customers experience a company. Their decisions carry the brand in every interaction. The task now is to enable them with AI and edge solutions that make those decisions informed, timely, and effective. #leadership #ai Andrew Chrostowski
“I Promise You, You Will Have Work to Do” - Nvidia CEO Jensen Huang Urges Everyone to Use AI as...
Nvidia CEO Jensen Huang told employees to automate every feasible task with AI, branding any effort to curb usage as "insane." His directive came after the company posted another record quarter, yet the stock fell amid investor doubts about sustained...
Claude Opus 4.5 Becomes a True Writing Partner
I have nothing fancy to say about Claude Opus 4.5. Only wow. It just saved me hours on client reports and it’s the first model to REALLY get the nuance of my writing style and voice and spar with me...
Define Your Win Function Before Letting AI Optimize
AI can optimize for almost anything. The question is: do you know what you want it to optimize for? Say you're using AI to design a new flight path. You could measure speed, on-time arrival, and fuel efficiency. But what...
Construction Workers Are Cashing in on the AI Boom
The rapid AI boom is driving a massive data‑center construction surge, prompting tech giants to pour billions into new facilities. Workers moving into these projects are seeing 25‑30% wage increases, with some supervisors earning over $100,000 and specialists topping $200,000....
DeepSeek Just Dropped Two Insanely Powerful AI Models that Rival GPT-5 and They're Totally Free
DeepSeek, a Chinese AI startup, unveiled two 685‑billion‑parameter models—DeepSeek‑V3.2 and the high‑performance DeepSeek‑V3.2‑Speciale—under an MIT open‑source license. The models employ a novel Sparse Attention architecture that halves inference costs for long‑context tasks, supporting 128,000‑token windows at roughly $0.70 per million...
Boardy AI Instantly Interviews Thousands, Reshaping Recruitment
I just applied. Boardy is the most disruptive thing I have seen in years. He (an AI) called me instantly after applying and we talked for half an hour. It can do that with thousands of applicants all at the same time. It...
AlphaFold Revolutionizes Biological and Biomedical Research
Great article from @jeremyakahn in Fortune on the huge impact of AlphaFold on biological and biomedical research.
Responsible AI Center to Combine Research With Industry Know-How
A coalition of universities and industry launched the Center on Responsible Artificial Intelligence and Governance (CRAIG), funded by the U.S. National Science Foundation, to develop methods, tools, and standards for ethical AI deployment. Faculty from Ohio State, Northeastern, Baylor and...
LeRobotHF's Breakthrough Sparks Collaborative Open AI Robotics
Remi and Simon created @LeRobotHF at HF which is having an amazing impact on the field so can’t wait to see what they can do with Uma! Let’s go collaborative and open AI robotics! https://t.co/9ezbLLuf5Z
ChatGPT Now Includes Interactive Finance Widget
just noticed this interactive "finance-widget" in chatgpt for the first time https://t.co/DAk0SqRcC2
MIT Offshoot Liquid AI Releases Blueprint for Enterprise-Grade Small-Model Training
Liquid AI, an MIT spin‑off, has published a 51‑page technical report that details the architecture, training curriculum, and post‑training pipeline behind its LFM2 family of small, on‑device foundation models. The report reveals a hardware‑in‑the‑loop architecture search that favors gated short...
AI Reviewer Beats NeurIPS Submissions, Signaling Future Impact
NeurIPS received 21,575 paper submissions this year. Our Agentic Reviewer, released last week, just surpassed this in number of papers submitted and reviewed. It's clear agentic paper reviewing is here to stay and will be impactful!
AI Slide Makeover Falls Short of True Design
i'm not too impressed with the "beautify this slide" result in google slides (before, after) makes me wonder what "beautiful" slides were in the training data .... https://t.co/iAsq8g3lid
Your AI-Generated Image of a Cat Riding a Banana Exists because of Children Clawing Through the Dirt for Toxic Elements....
The article highlights the hidden environmental and social toll of large language models, tracing their supply chain from child‑labour‑driven rare‑earth mining in the Congo to energy‑hungry data centres built in water‑scarce regions. It notes that training data curation often exposes...
DeepMind Hires Scientist to Explore Post‑AGI Future
Google DeepMind: “We are seeking a Research Scientist to explore the profound impact of what comes after AGI. Key responsibilities include defining critical research questions within these domains, collaborating with cross-functional teams to develop innovative solutions, and conducting experiments to...
Intellect-3 Beats Larger Models with 34% Gain
If the initial benchmarks scores (and graphs used for PR) showcased a 3x reduction in size for the same performance, I think the broader public reception would have been less tepid. Just looking at this, it just seems 7% behind other...
AI Models Block 87% of Single Attacks, but Just 8% when Attackers Persist
Cisco’s AI Threat Research team discovered that open‑weight large language models block 87% of single‑turn malicious prompts but see attack success soar to 92% when adversaries persist across multiple turns. The study evaluated eight popular models and found multi‑turn success...
Free GitHub README Outshines $999 AI Courses
Wild that a free GitHub README beats most $999 AI courses. Do yourself a favor: bookmark it. https://t.co/FgFzyJ4Mis
Transformers V5 RC Launches with Seamless Ecosystem Interoperability
Today we release the transformers version 5 RC! 🤗 With this, we enable e2e interoperability with our friends in ecosystem, ease up adding new models and simplify the library 🙌🏻 Read our blog to learn more: https://t.co/ysZW3btRgR https://t.co/PVKFRsH9Z2
The State of AI: Welcome to the Economic Singularity
Generative AI adoption remains highly uneven, with 95% of projects failing to generate returns while coding assistants promise transformative gains. Experts cite a classic productivity paradox, suggesting AI’s economic impact will follow a J‑curve that requires new data infrastructure and...
Open‑source AI “Computer Use Model” Speeds up Tasks
This AI controls your computer but does it a lot faster than previous attempts. It's called a "Computer Use Model." They are building an open source ecosystem around this. "Hey AI can you help me with my spreadsheets?"
Ditto Powers New Functional Vibe Coding App
okay fine this is fun someone built a working vibe coding app on top of my earlier project called ditto (self-building coding agent w ~500 lines of code) https://t.co/VK2yyTzUnS https://t.co/GOoqgreIgr
Runway Says Its New Text-to-Video AI Generator Has ‘Unprecedented’ Accuracy
Runway unveiled its Gen-4.5 text-to-video model, touting unprecedented physical accuracy and visual precision. The new system delivers more realistic motion, weight, and fluid dynamics while better adhering to detailed prompts. Gen-4.5 is being rolled out gradually to all users, maintaining...
Consistent ML Workflow Beats Magic Tricks
ML isn’t magic — it’s a workflow. 🧠⚙️ 1. understand data 2. choose right algorithm 3. train 4. test 5. optimize 6. deploy + monitor + retrain The winners are the teams who run this loop consistently. #MachineLearning #AI #DataScience #MLops https://t.co/LxCzFgQFe5
RebuilderAI Secures Two CES 2026 Innovation Awards
RebuilderAI Wins Two CES 2026 Innovation Awards https://t.co/foF6ZlFHDF #CES2026 #AI #IoT #5G
AWS Re:Invent 2025 - All the News and Updates as It Happens
AWS re:Invent 2025 kicked off in Las Vegas with live coverage highlighting three marquee keynotes. CEO Matt Garman opened the event outlining Amazon’s multi‑year cloud strategy, followed by AI and Data VP Dr. Swami Sivasubramanian unveiling the next generation of...
StreamUnlimited Debuts Custom Voice LLM at CES 2026
StreamUnlimited Launches Customizable Voice LLM Reference Integration for Audio Agent Products at CES 2026 https://t.co/00kzTGxHWp @CES #AI #IoT #CES2026 https://t.co/tgGOHVH49w