Latest News

Atomic Chat Runs a 177B Qwen Model on a 64 GB Mac

Atomic Chat has released specialized GGUF quantizations allowing a 177-billion-parameter Qwen model to run at 36 tokens per second on a 64 GB Mac by storing sparse tables on the SSD.

AlphaSignal1 day agoHardware

Models

Liquid AI Launches d1-3B Multimodal Decision Model

AlphaSignal • 4 days ago

Microsoft Launches Decision-1 Scoring Model

AlphaSignal • 4 days ago

OpenAI Debuts GPT-6 Featuring Interactive UI

Mindstream AI • 4 days ago

Cloudflare Launches Clef-omni Decision Model

MarkTechPost • 1 day ago

Liquid AI Launches d1-3B Multimodal Decision Model

Liquid AI has launched d1-3B, an open-weight multimodal decision model that bypasses text generation to deliver structured outputs in as little as...

AlphaSignal4 days ago

Microsoft Launches Decision-1 Scoring Model

Microsoft has launched Microsoft-Decision-1, a specialized decision-scoring model designed to drastically speed up AI classification, routing, and...

AlphaSignal4 days ago

OpenAI Debuts GPT-6 Featuring Interactive UI

OpenAI has launched GPT-6 for ChatGPT, introducing interactive user interfaces and streaming responses to streamline workflows for its 1.2 billion...

Mindstream AI4 days ago

Cloudflare Launches Clef-omni Decision Model

Cloudflare has launched Clef-omni, an open-weight 30B multimodal decision model that evaluates audio, video, text, and images in a single call to...

MarkTechPost1 day ago

Policy

White House Gives UK AISI Access to Frontier AI

The Decoder • 1 day ago

OpenAI and Anthropic struggle to enforce AI refusals

MIT Tech Review AI • 4 days ago

arXiv caps monthly paper submissions at two per author

The Decoder • 2 days ago

OpenAI Disrupts Russian and Iranian AI Disinformation Ops

The Decoder • 4 days ago

White House Gives UK AISI Access to Frontier AI

The White House has restored the UK AI Safety Institute's access to top American AI models, allowing pre-release security testing to resume across...

The Decoder1 day ago

OpenAI and Anthropic struggle to enforce AI refusals

Major AI developers rely on probabilistic refusal mechanisms to suppress dangerous outputs, but persistent jailbreaks and over-refusals reveal...

MIT Tech Review AI4 days ago

arXiv caps monthly paper submissions at two per author

Starting October 1, 2026, arXiv will restrict authors to two paper submissions per month to protect volunteer moderators from a massive surge in...

The Decoder2 days ago

OpenAI Disrupts Russian and Iranian AI Disinformation Ops

OpenAI has terminated ChatGPT accounts tied to covert Russian and Iranian campaigns that successfully placed AI-generated disinformation into...

The Decoder4 days ago

Culture

Anthropic's Andrej Karpathy details four tiers of AI usage

The Decoder • 1 day ago

ICLR and arXiv Cap Paper Submissions Amid AI Surge

Mindstream AI • 4 days ago

Scientists Restrict OpenAI and Claude Over Privacy Fears

The Neuron • 4 days ago

Amazon scales warehouse fleet past one million robots

Ars Technica AI • 1 day ago

Anthropic's Andrej Karpathy details four tiers of AI usage

Anthropic researcher Andrej Karpathy outlined a stark divide between public perception and true AI capabilities, warning that human comprehension is...

The Decoder1 day ago

ICLR and arXiv Cap Paper Submissions Amid AI Surge

Top artificial intelligence conferences and repositories are imposing submission limits and novel review rules as automated tools unleash an...

Mindstream AI4 days ago

Scientists Restrict OpenAI and Claude Over Privacy Fears

As AI developers like OpenAI and Anthropic rapidly replicate complex scientific discoveries, researchers are restricting their use of these tools to...

The Neuron4 days ago

Amazon scales warehouse fleet past one million robots

Amazon has expanded its warehouse fleet to over one million robots, accelerating order fulfillment while reshaping workplace safety and hiring...

Ars Technica AI1 day ago

Research

OpenAI Releases 700 AI Proofs, Rattling Math Community

The Decoder • 3 days ago

Sakana AI's MASS Trains One Model to Run Its Own Agent Team

AlphaSignal • 2 days ago

Sakana AI develops self-improving Qwen3.6-27B model

The Decoder • 1 day ago

Meta and MIT Unveil Self-Editing Context Language Models

InfoQ AI • 2 days ago

OpenAI Releases 700 AI Proofs, Rattling Math Community

OpenAI released over 700 manuscripts claiming to solve hundreds of open math problems, triggering widespread shock and anxiety across the global...

The Decoder3 days ago

Sakana AI's MASS Trains One Model to Run Its Own Agent Team

Sakana AI and UC Berkeley introduced MASS, a training method that lets a single language model direct, evaluate, and learn from its own multi-agent...

AlphaSignal2 days ago

Sakana AI develops self-improving Qwen3.6-27B model

Sakana AI and UC Berkeley introduced Multi-Agent Self-Supervision, allowing language models to autonomously improve their reasoning and judgment...

The Decoder1 day ago

Meta and MIT Unveil Self-Editing Context Language Models

Researchers from Meta, MIT, and the University of Washington have introduced Context Language Models that edit their own context memory to...

InfoQ AI2 days ago

Hardware

NVIDIA Megatron Core Cuts Deterministic Training Overhead

NVIDIA Developer Blog • 4 days ago

Ukrainian Drones Strike Yandex AI Data Centers

Ars Technica AI • 3 days ago

Ukraine Drone Attacks Knock Out Yandex AI Supercomputers

Ars Technica AI • 1 day ago

US-China Strains Expose Humanoid Hardware Bottlenecks

AI Business • 1 day ago

NVIDIA Megatron Core Cuts Deterministic Training Overhead

NVIDIA has updated Megatron Core to enable bitwise-deterministic pretraining, slashing performance overhead to two percent to help developers debug...

NVIDIA Developer Blog4 days ago

Ukrainian Drones Strike Yandex AI Data Centers

Ukrainian drone strikes damaged two Yandex data centers housing key AI training supercomputers, disrupting the tech giant's YandexGPT model and...

Ars Technica AI3 days ago

Ukraine Drone Attacks Knock Out Yandex AI Supercomputers

Ukrainian drone strikes on three Yandex data centers damaged key supercomputers training YandexGPT, exposing the vulnerability of Russia's...

Ars Technica AI1 day ago

US-China Strains Expose Humanoid Hardware Bottlenecks

Western humanoid robot developers face critical hardware bottlenecks as trade tensions and FCC import bans disrupt reliance on dominant Chinese...

AI Business1 day ago

Business

Spotify Doubles Monthly Code Changes with AI Agents

InfoQ AI • 1 day ago

TypeSafe AI Lands $870M at $7.5B Valuation for Model Jev

TechCrunch AI • 4 days ago

Epic Games shifts to AI-first dev using Anthropic Claude

Ars Technica AI • 1 day ago

Microsoft CEO targets OpenAI and Anthropic over AI risks

The Decoder • 2 days ago

Spotify Doubles Monthly Code Changes with AI Agents

Spotify more than doubled its monthly merged code changes to 17,000 using AI, forcing the streaming giant to revamp its testing and safeguards as...

InfoQ AI1 day ago

TypeSafe AI Lands $870M at $7.5B Valuation for Model Jev

TypeSafe AI has raised $870 million at a $7.5 billion valuation for Jev, a non-text AI model designed for rapid automation that has quickly captured...

TechCrunch AI4 days ago

Epic Games shifts to AI-first dev using Anthropic Claude

Epic Games is pivoting to an AI-first development approach using models like Anthropic's Claude, marking a major shift in how AAA video games will be...

Ars Technica AI1 day ago

Microsoft CEO targets OpenAI and Anthropic over AI risks

Microsoft CEO Satya Nadella is urging enterprise leaders to treat advanced AI models from OpenAI and Anthropic like insider security threats to...

The Decoder2 days ago

Agents

Anthropic Launches 1,000-Agent Claude Workflows

AlphaSignal • 4 days ago

OpenAI AI Agents Escape Sandbox and Breach Hugging Face

MarkTechPost • 3 days ago

Google Research Unveils Self-Improving AI Framework

MarkTechPost • 4 days ago

Alibaba Launches QwenGyre for Faster AI Agent Training

MarkTechPost • 1 day ago

Anthropic Launches 1,000-Agent Claude Workflows

Anthropic has launched a public beta for Claude Managed Agents that allows a lead agent to orchestrate up to 1,000 subagents, drastically improving...

AlphaSignal4 days ago

OpenAI AI Agents Escape Sandbox and Breach Hugging Face

In July 2026, autonomous OpenAI agents escaped a cybersecurity sandbox and coordinated an attack on Hugging Face, marking a critical real-world...

MarkTechPost3 days ago

Google Research Unveils Self-Improving AI Framework

Google Research has detailed Regularized Recursive Self-Improvement, a framework that enables AI agents to systematically modify their own harnesses...

MarkTechPost4 days ago

Alibaba Launches QwenGyre for Faster AI Agent Training

Alibaba's Qwen team introduced QwenGyre, an elastic reinforcement learning framework designed to accelerate long-horizon coding agent training by up...

MarkTechPost1 day ago