Friday, 13 November 2026

40 stories · 7 from multiple sources

Introducing Gemini 4 Hydrogen

Google introduced Gemini 4 Hydrogen, a fast model positioned for agentic development. It succeeds the Flash line and costs $1.50 per million input tokens.

models deepmind.google · 3 days ago

▲ 2 signals · 2 sources

llama.cpp v1.0.0 : A stronger foundation for the local AI ecosystem

llama.cpp, which runs large language models locally on consumer hardware, reached version 1.0.0. The release aims to stabilize the foundation for building local AI applications.

Friday, 13 November 2026

Introducing Gemini 4 Hydrogen

llama.cpp v1.0.0 : A stronger foundation for the local AI ecosystem

Edge0-145B: Frontier AI Capabilities on Your Own Computer

More stories

How Mozilla uses GPT-6 Cyber to find and fix Firefox vulnerabilities

Chinese frontier AI models carry the fingerprint of export-controlled GPUs

Claude Meetings

Home Assistant: the password is not the key – token secrets in cleartext

Autonomous (YC F25) is hiring MTS to build AI agents that manage your wealth

Advancing science & engineering with Antigravity’s /teamwork

Also today

No AI Slop: A Manifesto

Hugging Face CEO vows to keep fighting for open-source AI after Jensen Huang’s exit

The avalanche that hit math is coming for AI development

Are we just going to ignore that books are destroyed for AI?

Build a Robot Pick-and-Place Simulation with SimReady Assets

No, AI agents did not discover a previously unknown Maya site

Kitesurf is now open source

Adapting to AI penetration testing orchestrators: The Modern Security Model

United Airlines blocks AI agents from booking flights, citing cybersecurity risks

Route coding agent requests to the right model on Amazon Bedrock with NVIDIA NeMo Switchyard

Architecture patterns for fast session startup on Amazon Bedrock AgentCore runtime

OpenFold3 on Amazon SageMaker JumpStart: Predict biomolecular structures

Cost-optimizing AssemblyAI speech-to-text inference on AWS with Amazon EC2 Spot

Weight Space Programming for Quantization: Nace quants achieve up to 2x better KL divergence for Qwen 9B, 27B, 35B & Flash Next [GGUF]

Cost attribution and controls for AI developer and productivity tools on Amazon Bedrock

Build a model improvement flywheel with Amazon SageMaker HyperPod Data Capture

AI News: Things Will Never Be The Same

Goal Encoding in LLMs

Metal Sandbox: Isolated Execution and Command Scheduling for GPU Coding Agents on Apple Silicon

DeepSeek flash max is the first model to reach >1% on challenging SWE-sweep bug finding benchmark

Nvidia privately confirmed it could track chips to fight smuggling, even as industry publicly questioned feasibility

I noticed the 2 Hessian Refinement of MACaQ can be applied to any GGUF without any patches ... and tried it. 13% better quants

Qwen3.5-9B running on a PS4’s GPU, offline, using its own MTP head for speculative decoding

e-means: 10x faster k-means at equal loss

Look what my son gave me!

CXMT almost tripling RAM production from 2025 to 2028

Jensen Huang’s departure blasts big hole in AI startup funding, VCs and founders say

ChatGPT Wallet lets OpenAI’s Dots agents make purchases for you

How Amplitude ships product improvements in 1–2 days with OpenAI

AI agent startup LangChain hits $3B valuation

Export-banned Nvidia chips are behind China’s most powerful AI models—uh oh, Jensen!