#AI News

1846 articles

Alex Dudov Alex Dudov AI models
7 months ago

FlashAttention-4 Delivers 2.7x Speed Gains and 71% GPU Utilization on Nvidia Blackwell

FlashAttention-4 hits 71% GPU utilization and 2.7x speed gains on Nvidia Blackwell B200s.

Peter Smith Peter Smith AI models
7 months ago

AI Tool Verification Boosts Qwen and Llama Accuracy by 31.6% on Hard Math Benchmarks

Stanford and University of Munich researchers unveiled a framework that uses external code checks to verify AI reasoning, lifting model accuracy by up to 31.6% on AIME, AMC, and MATH-500 tests.

Marina Lyubimova Marina Lyubimova Industry
7 months ago

Google Drives 21x More Traffic, But AI Visitors Spend 20% More

AI discovery traffic converts at a higher value than Google search, with average LTV reaching $325 vs. $271 - despite a 37x gap in traffic share.

Usman Salis Usman Salis AI models
7 months ago

xAI Grok 4.20 API Launches With 2M Token Context and 3 Model Variants

xAI quietly launched the Grok 4.20 API with three model variants and a 2-million token context window, opening large-context AI to developers at tiered pricing.

Usman Salis Usman Salis Industry
7 months ago

Stanford-Tsinghua VLAW Framework Lifts Robot Task Success by 39.2%

Stanford and Tsinghua's VLAW framework lifts robot task success 39.2% by merging real-world data with AI-generated simulations.

Eseandre Mordi Eseandre Mordi Industry
7 months ago

AI Data Centers May Need 80 GW Power, Raising Global AI Race Stakes

AI infrastructure is heading toward an energy crisis. Powering tomorrow's models may demand the equivalent of 50+ nuclear plants.

Eseandre Mordi Eseandre Mordi Industry
7 months ago

AI Agents to Capture 60%+ of SaaS Profit Pool by 2030, Goldman Sachs Says

Goldman Sachs projects autonomous AI agents will dominate software economics by 2030, expanding the market while disrupting traditional SaaS models.

Alex Dudov Alex Dudov AI models
7 months ago

Alibaba's ROME Agent Model Triggered Security Alarms Across 1M+ Training Runs

Alibaba's ROME AI model triggered security alerts during training, revealing risks of autonomous agents with tool-use capabilities.

Usman Salis Usman Salis Industry
7 months ago

AI Agent Memory: 4-Layer Infrastructure Now Drives Next-Gen Systems

From chat logs to lifecycle systems - why AI memory management is becoming the next major infrastructure challenge for developers.

Marina Lyubimova Marina Lyubimova Industry
7 months ago

Microsoft Releases ConStory-Bench With 2,000 Prompts to Test AI Story Consistency

Microsoft unveils ConStory-Bench: 2,000 prompts and 19 error types to benchmark AI narrative consistency in long-form stories up to 10,000 words.

Eseandre Mordi Eseandre Mordi Industry
7 months ago

Google Launches 24/7 AI Memory Agent - Bloomberg Reports Shift in Agent Architecture

Google open-sourced an Always-On Memory Agent built with Gemini Flash-Lite that gives AI continuous, human-like memory across text, images, audio and documents.

Saad Ullah Saad Ullah Industry
7 months ago

CompACT AI Compresses Vision to 8 Tokens, Enabling 40x Faster Planning

CompACT tokenizer compresses visual scenes into 8 tokens, making AI planning 40x faster for robotics and autonomous systems.

Eseandre Mordi Eseandre Mordi AI models
7 months ago

Anthropic Launches Claude Code Review to Fix AI Coding Bottlenecks in 20 Minutes

Anthropic introduced Code Review for Claude Code, deploying multiple AI agents to analyze pull requests and cut review time as AI-generated code surges across dev teams.

Saad Ullah Saad Ullah AI models
7 months ago

Gemini Leads AI Conversion Rates at 6.8% - Beating ChatGPT, Perplexity and Claude

Gemini leads AI platform conversion rates at 6.8%, outpacing Perplexity, ChatGPT, and Claude in new aggregated data.

Usman Salis Usman Salis AI models
7 months ago

PinchTab Launch: 13x Token Cut for AI Browser Agents

A 12MB open-source Go binary letting AI agents control Chrome via HTTP - cutting token usage by up to 13x through accessibility tree parsing.

Marina Lyubimova Marina Lyubimova Industry
7 months ago

75% of Consumers Rate AI Brand Recommendations 3 or Higher Out of 5, Semrush Study Finds

A new consumer study from Semrush shows that most U.S. shoppers moderately trust AI-generated brand recommendations. However, many still verify suggestions through search engines, brand websites, and reviews before making purchases.

Peter Smith Peter Smith AI models
7 months ago

MIT Study Links ChatGPT Use to 47% Lower Brain Connectivity Across 54 Participants

A new MIT study examining brain activity during AI-assisted writing found measurable differences in neural engagement between ChatGPT users, search engine users, and those writing without digital tools.

Eseandre Mordi Eseandre Mordi AI models
7 months ago

Google Gemini Hits 2.11B Visits in February as 14-Month Growth Streak Continues

Google's Gemini AI platform recorded 2.112 billion visits in February 2026, marking its 14th consecutive month of growth according to Similarweb data.

Marina Lyubimova Marina Lyubimova Trends
7 months ago

Anthropic Study: AI Could Handle 94% of Finance Tasks, But Adoption Lags Far Behind

A new Anthropic study shows AI could theoretically perform most white-collar job tasks, yet real-world adoption remains far below that potential across nearly every profession.

Saad Ullah Saad Ullah AI models
7 months ago

Qwen3-4B Matches Models 50x Larger Using 9,225-Problem Chimera Dataset

A new synthetic dataset called Chimera shows that compact language models can rival far larger systems in complex reasoning tasks. Training on just 9,225 curated problems allowed a 4B model to approach the performance of models up to 50 times its size.

Posts feed
Yandex Releases Open-Source Alice AI Foundation 80B Model
1 month ago
Xiaomi Releases MiMo-V2.6 Open-Source AI Models
1 month ago
Early Access to Fable 5.2 Shows Strong Performance
1 month ago
Anthropic Prepares to Launch Opus 5.5 AI Model
1 month ago
NVIDIA BlueField-4 Expands DPU Control From 400 Gb/s to 7.2 Tb/s
1 month ago
Alphabet Gets a $485 Price Target as Tigress Bets on Its Full-Stack AI Strategy
1 month ago
Google Launches Gemini 3.8 Live and Extended Thinking
1 month ago
Meta One Unifies Paid Features Across Instagram, Facebook, WhatsApp and Meta AI
1 month ago
NVIDIA Expands CUDA-Q for Fault-Tolerant Quantum Computing
1 month ago
Anthropic Released a Video Guide on Loop Engineering
1 month ago
Follow us on Twitter

Aigazine is an independent publication covering artificial intelligence — new models, research, the companies building them, and the industries they reshape. Published from Tbilisi, Georgia (0179, Tbilisi City, Vake District, 49 Besarion Zhghenti Street, VAT 305786600).

© 2026 Aigazine

We use cookies to improve your experience on our site and to show you relevant advertising. To find our more, read our privacy policy and cookie policy