Artificial Intelligence
The latest breakthroughs in AI, machine learning, and neural networks.
64 articles
Google Gemma 4 Comes to Android: On-Device AI in 140+ Languages, No Cloud Required
Google's AICore Developer Preview brings Gemma 4 natively to Android devices — offline, privacy-preserving AI inference in over 140 languages that upgrades automatically to Gemini Nano 4.
LG Releases EXAONE 4.5: Open-Source Vision-Language AI That Outscores GPT-5-mini
LG AI Research's EXAONE 4.5 is a 33B multimodal VLM with Hybrid Attention architecture that outscores GPT-5-mini and Claude 4.5 Sonnet on STEM benchmarks — and it's fully open-source.
Tufts Researchers Build AI That Uses 1% of the Energy and Outperforms Neural Nets
A Tufts University neuro-symbolic AI achieved 95% accuracy on complex reasoning tasks while consuming just 1% of the energy of conventional deep learning systems.
NVIDIA's AI-Q Blueprint Brings Enterprise Agentic AI to Adobe, Salesforce, and SAP
NVIDIA's AI-Q Blueprint gives enterprises an open framework for building AI agents that perceive, reason, and act — slashing query costs by 50% with a hybrid routing architecture.
PrismML's Bonsai Is a 1-Bit LLM That Runs on a Smartphone and Matches Full-Size Models
Caltech startup PrismML emerged from stealth with Bonsai, a 1-bit LLM family that's 14x smaller, 8x faster, and 5x more energy-efficient than standard 8B models — and runs on an iPhone.
Meta Launches Llama 4 Scout and Maverick: Multimodal MoE AI Goes Open-Weight
Meta's Llama 4 Scout and Maverick bring multimodal mixture-of-experts AI to the open-source community, with an unprecedented 10 million token context window.
Microsoft Launches MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2
Microsoft unveiled MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2 on April 2 — three new in-house foundation models now available in Microsoft Foundry.
Alibaba's Qwen3.6-Plus Delivers 1M-Token Context and Repository-Level Agentic Coding
Qwen3.6-Plus arrives with a default 1 million-token context window and breakthrough agentic coding performance, enabling AI that can navigate and rewrite entire software repositories autonomously.
Google Gemma 4 Launches With Four Sizes, Apache 2.0 License, and a Top-3 Open Model Ranking
Google's Gemma 4 arrives with model sizes from 2B to 31B, a permissive Apache 2.0 license, native multimodal support across all sizes, and the #3 spot on the global open model leaderboard.
OpenClaw Becomes GitHub's Most-Starred Project as the Open-Source AI Agent Wave Crests
The open-source AI agent framework OpenClaw has surpassed 250,000 GitHub stars, overtaking React — marking a pivotal shift in how developers worldwide build and ship autonomous AI agents.
Google's TurboQuant Compresses AI Memory 6x With Zero Accuracy Loss
Google Research's TurboQuant cuts LLM KV cache memory to 3 bits without accuracy loss, delivering up to 8x inference speedups on NVIDIA H100 GPUs — with no retraining required.
Mistral Voxtral TTS Is an Open-Weight Voice AI That Rivals ElevenLabs
Mistral AI's new 4B open-weight TTS model supports zero-shot voice cloning in 9 languages with 70ms latency — and the weights are free to download.
Google Opens Gemini 3 Deep Think to API Developers and Ultra Subscribers
Google's frontier reasoning model Gemini 3 Deep Think is now live for Ultra subscribers and API developers, built for scientific research, math, and complex engineering.
NVIDIA Unveils Vera Rubin GPU Architecture and Open AI Agent Platform at GTC 2026
NVIDIA's GTC 2026 brings the Vera Rubin GPU with 3-4x Blackwell AI compute density, the Nemotron 3 Super 120B coding model, and an open-source platform for safe autonomous AI agents.
ElevenLabs and IBM Give Enterprise AI Agents a Human Voice
ElevenLabs integrates with IBM's watsonx Orchestrate, giving enterprise AI agents natural voice for richer, human-feeling agentic workflows at scale.
Google's Gemini 3.1 Flash Live Arrives With Real-Time Voice and ChatGPT Memory Import
Google's March Gemini Drop delivers real-time voice AI, ChatGPT history migration, and free Personal Intelligence — a major leap for Gemini's consumer platform.
Apple's iOS 27 Opens Siri to Rival AI Assistants Beyond ChatGPT
Apple's iOS 27 will let Siri hand off to third-party AI assistants, positioning iPhone as an open AI platform in its biggest intelligence overhaul yet.
Oracle AI Database 26ai Embeds Persistent Agent Memory Into the Database Engine
Oracle's AI Database 26ai embeds persistent agent memory inside the database engine, unlocking a new class of data-centric AI agents for enterprise deployments.
Cursor Composer 2: Frontier AI Coding at 86% Lower Cost
Cursor launches Composer 2, a code-only AI agent with a 200K token context window, 73.7 SWE-bench score, and 86% lower cost than its predecessor.
Mistral Forge Lets Enterprises Train Custom AI Models on Their Own Data
Mistral's new Forge platform gives enterprises end-to-end custom model training using open-weight models including the new 119B Mistral Small 4, backed by NVIDIA at GTC 2026.
Google DeepMind Partners With Agile Robots to Bring Gemini AI to Commercial Humanoids
Google DeepMind's Gemini Robotics foundation models are coming to Agile Robots' commercial humanoid hardware in a March 2026 partnership expanding physical AI deployment.
Figma Opens Design Canvas to AI Agents via MCP Integration
Figma launches MCP server expansion enabling AI coding agents to generate and modify design assets directly on live canvases with a new skills system.
Xiaomi Pledges $8.7B AI Investment, Launches MiMo-V2-Pro Rivaling Frontier Models
Xiaomi CEO Lei Jun commits 60B yuan to AI over three years and unveils MiMo-V2-Pro, a 1T parameter reasoning LLM approaching frontier benchmarks at a fraction of the cost.
Anthropic Launches Claude Code Channels for Telegram and Discord
Developers can now control a local Claude Code session from Telegram or Discord, sending messages that flow through chat apps to a local instance with full filesystem access.
MIT Researchers Develop a New Metric That Catches Overconfident AI Models Before They Hallucinate
A new uncertainty measurement technique from MIT combines self-consistency checks with cross-model disagreement to flag when LLMs generate confident but incorrect responses.
Microsoft Reshuffles Copilot AI Division — Suleyman Pivots to Model Building
Mustafa Suleyman shifts focus to next-gen AI model development as ex-Snap exec Jacob Andreou unifies consumer and enterprise Copilot under one roof.
Surgical Robots and Disney's Olaf Take the Stage at GTC 2026 — NVIDIA's Physical AI Push Goes From Factory to Operating Room
Johnson & Johnson MedTech, CMR Surgical, and Moon Surgical adopt NVIDIA's healthcare physical AI platform, while Disney's free-roaming Olaf robot steals the keynote spotlight.
NVIDIA Ships the DGX Station GB300 — 748GB of Coherent Memory, 20 Petaflops, and Trillion-Parameter Models at Your Desk
The first DGX Station GB300 systems arrive with the Grace Blackwell Ultra Desktop Superchip, while DGX Spark gains 4-unit clustering for desktop-scale AI data centers.
Five Automakers Are Building Level 4 Self-Driving Vehicles on NVIDIA Drive Hyperion — From Nissan to BYD
NVIDIA reveals that Nissan, BYD, Geely, Isuzu, and Hyundai are developing Level 4 autonomous vehicles using its Drive Hyperion platform, expanding AI chips beyond the data center.
NVIDIA GTC 2026 — Jensen Huang Unveils Vera Rubin GPUs, Groq 3 LPU, Kyber Racks, and the First Data Center in Space
NVIDIA's GTC 2026 keynote reveals the Vera Rubin GPU platform with 10x performance per watt, the Groq 3 LPU for 35x inference gains, and a $1 trillion Blackwell-to-Rubin opportunity.
OpenAI Launches GPT-5.4 — A Million-Token Context Window and Native Computer Use Arrive Together
GPT-5.4 ships with a 1M-token API context window, built-in computer-use capabilities, and a new Thinking mode that scores at expert level on economic benchmarks.
Google Launches Gemini Embedding 2 — The First AI Model That Maps Text, Images, and Video Into a Single Search Space
Google's new natively multimodal embedding model jointly maps text, images, and video into a unified vector space, enabling cross-modal retrieval and RAG applications.
Perplexity Unveils 'Personal Computer' — A Mac Mini That Runs an Always-On AI Agent With Full Local Access
At its Ask 2026 conference, Perplexity launches Personal Computer software for Mac mini and expands Computer for Enterprise with SOC 2, SAML SSO, and Slack integration.
Yann LeCun's AMI Labs Raises $1.03 Billion in the Largest European Seed Round Ever — Betting on World Models Over LLMs
Turing Award winner Yann LeCun leaves Meta to build Advanced Machine Intelligence Labs, raising $1.03B at a $3.5B valuation to develop JEPA-based world models.
Netflix Acquires Ben Affleck's InterPositive for Up to $600 Million — The Biggest AI Filmmaking Deal Yet
Netflix snaps up the stealth AI post-production startup InterPositive, bringing its 16-person team in-house to build AI-powered tools for relighting, VFX, and continuity.
NVIDIA Launches Nemotron 3 Open Models at GDC — 120B Parameters With 5x the Throughput
NVIDIA's Nemotron 3 family ships in Nano, Super, and Ultra sizes with up to 1M-token context, already adopted by CrowdStrike, Cursor, Perplexity, and Zoom.
OpenAI Is Bringing Sora Video Generation Directly Into ChatGPT — Giving Hundreds of Millions of Users Access
According to The Information, OpenAI plans to embed Sora's video-generation capabilities into the ChatGPT interface, mirroring how DALL-E image creation was integrated.
Google Embeds Gemini Across Docs, Sheets, Slides, and Drive — Your Workspace Just Got an AI Co-Pilot
Google rolls out deep Gemini AI integration across its entire Workspace suite, including full-document generation in Docs, real-time data population in Sheets, and AI-powered Drive search.
Google Launches Gemini 3.1 Flash-Lite — A Lightning-Fast AI Model at Just $0.25 Per Million Tokens
Google DeepMind's new Flash-Lite model delivers 2.5x faster responses than its predecessor at a fraction of the cost, making production-scale AI deployment dramatically more affordable.
The Pentagon Just Labeled Anthropic a 'Supply Chain Risk' and Ordered a Federal Phase-Out of Claude
After Anthropic refused to remove guardrails banning mass surveillance and autonomous weapons, the Department of Defense triggered a six-month phase-out of Claude from classified networks.
OpenAI Launches GPT-5.4 With Native Computer Use, 1M Token Context, and 33% Fewer Errors
OpenAI's newest frontier model can autonomously operate your desktop, processes up to a million tokens of context, and cuts hallucinations by a third compared to GPT-5.2.
Jack Dorsey's Block Cuts Nearly 40% of Its Workforce — And Says AI Will Fill the Gaps
Block slashes its headcount from over 12,000 to roughly 7,500 employees, citing AI-driven productivity gains as a core reason the company can operate leaner.
Apple's M5 Pro and Max Chips Fuse Neural Accelerators Into Every GPU Core — Delivering 4x the AI Compute
Apple's new Fusion Architecture bonds two 3nm dies into a single SoC, embedding dedicated neural accelerators in every GPU core for massive on-device AI gains.
OpenAI Releases GPT-5.3 Instant — Cutting Hallucinations by 27 Percent and Delivering Snappier Answers
GPT-5.3 Instant focuses on reliability over raw power, reducing hallucination rates by nearly 27 percent while trimming unnecessary refusals and preambles.
Ayar Labs Raises $500 Million to Mass-Produce Optical AI Interconnects — Backed by NVIDIA and AMD
Ayar Labs closes a massive Series E to scale optical chiplets that replace copper connections with light, delivering up to 20x more bandwidth per watt for AI data centers.
DeepSeek Unveils V4 — A Trillion-Parameter Multimodal Model That Generates Text, Images, and Video
DeepSeek's V4 model enters the frontier tier with trillion-parameter multimodal capabilities spanning text, image, and video generation plus elite coding performance.
Atlassian Brings AI Agents Into Jira — Managed Alongside Human Workers on the Same Dashboard
Jira now lets teams assign tasks to AI agents from the same board used for human work, plus a new Rovo MCP Server connects Claude, Cursor, and Gemini to Atlassian data.
Anthropic Acquires Vercept to Supercharge Claude's Computer-Use and Visual AI Agent Capabilities
Anthropic's acquisition of Vercept brings specialized computer vision and UI understanding tech to Claude, accelerating the push toward fully autonomous AI agents.
Apple Is Rebuilding Siri From the Ground Up With LLM-Powered Conversational Intelligence in iOS 26.4
Apple's Siri overhaul replaces the command-based architecture with on-device large language models, bringing natural conversation and app-aware context to one billion iPhones.
Anthropic Embeds Claude Directly Into Excel, PowerPoint, and Slack With New Enterprise Plugins
Claude's Cowork Enterprise Plugins let AI edit spreadsheets, build presentations, and pass context across Office apps — with prebuilt vertical integrations.
AT&T Slashes AI Costs by 90 Percent Using Small Language Models That Process 27 Billion Tokens Daily
AT&T's multi-agent architecture routes tasks to specialized small models, cutting AI infrastructure costs while scaling to 27 billion tokens per day.
NVIDIA Debuts Nemotron 3 — Open Models With 4x Throughput and a Million-Token Context Window
NVIDIA's Nemotron 3 family ships three tiers of open models optimized for agentic AI, plus 3 trillion tokens of training data for the community.
Perplexity Launches Computer — An AI Agent That Orchestrates 19 Models to Tackle Complex Workflows
Perplexity's new Computer platform routes tasks across Claude, Gemini, GPT-5.2, and Grok, ushering in a multi-model orchestration era for AI agents.
HyperNova 60B Uses Quantum-Inspired Math to Halve an LLM's Size With Near-Zero Accuracy Loss
Multiverse Computing's free HyperNova 60B compresses a 120B-parameter model by 50% using quantum tensor methods, benchmarking 5x better on tool-calling tasks.
ServiceNow's New AI Agents Already Resolve 90 Percent of Its Own IT Requests Autonomously
ServiceNow launches Autonomous Workforce with AI specialists that handle Level 1 IT support, processing requests 99% faster than human agents.
Inception Labs Launches Mercury 2 — The First Reasoning LLM Built on Diffusion Architecture
Mercury 2 processes tokens in parallel via iterative denoising, hitting 1,000 tokens per second while matching top reasoning models on benchmarks.
MIT Researchers Develop a Proxy Model Technique That Doubles LLM Training Speed
A new MIT method uses a lightweight proxy model to predict reasoning outputs, cutting the reinforcement learning rollout bottleneck in half.
New Self-Distillation Technique Triples LLM Inference Speed With a Single Model
Researchers achieve 3x faster LLM inference by baking multi-token prediction directly into model weights — no draft model or extra hardware required.
Fei-Fei Li’s World Labs Raises $1 Billion to Build AI That Understands the Physical World
World Labs secures $1B from Nvidia, AMD, Autodesk, and Fidelity to scale its MARBLE spatial intelligence platform for gaming, VFX, and robotics.
Google Gemini 3.1 Pro Doubles Reasoning Performance With a New Three-Tier Thinking System
Google DeepMind’s Gemini 3.1 Pro scores 77.1% on ARC-AGI-2, more than doubling its predecessor’s reasoning with a three-tier thinking architecture.
OpenAI Teams Up With McKinsey, BCG, Accenture, and Capgemini to Bring AI Agents to the Enterprise
OpenAI launches Frontier Alliances with McKinsey, BCG, Accenture, and Capgemini to deploy AI agents across enterprise workflows, marking a major step for agentic AI in business.
Hugging Face Acquires ggml.ai, Giving llama.cpp a Permanent Open-Source Home
Hugging Face acquires ggml.ai, bringing llama.cpp and the GGUF model format under its umbrella while keeping everything MIT-licensed and open-source for local AI inference.
AI System Discovers 25 New Magnetic Materials That Could Replace Rare Earth Elements
University of New Hampshire researchers used AI to scan 67,000+ compounds and found 25 high-temperature magnets that could reshape the clean energy supply chain.
Model Context Protocol Goes Mainstream as OpenAI, Google, and Microsoft Adopt Anthropic's Standard
The protocol that lets AI systems connect to databases and tools is now governed by the Linux Foundation, with every major AI lab on board.