
California Sues OpenAI Over A.I. Safety Issues
The state attorney general filed a lawsuit accusing the company of failing to put appropriate safeguards on its artificial intelligence and putting the public at risk.
The daily dispatch from the AI arms race. 783 stories from 130 feeds, ranked by heat and rebuilt every morning.
Everything in the issue, ranked by heat.
The stories with the most heat right now, plus one pick from every corner of the AI world.

The state attorney general filed a lawsuit accusing the company of failing to put appropriate safeguards on its artificial intelligence and putting the public at risk.
An open source design system that's fully customizable and agent ready
text-to-image · license other
Agentic robotics, centered on an Embodied System 2 formed by a foundation model and embodied harness, requires a shared evaluation framework and research roadmap to achieve reliable physical autonomy.

Design the system, not every output.
A hands-on guide to training Mixture-of-Experts models
Perplexity crossed a $1B annual revenue run rate today , CEO Aravind Srinivas told Semafor at a live event. The Information reported previously under $250M at the start of the year and over $750M by August, with Computer driving some of the growth. At usual…
The newest videos and Instagram posts from the AI accounts worth following, refreshed every morning.






















No new upload this week: Andrej Karpathy, Yannic Kilcher, AI Jason, MattVidPro, Anthropic, 3Blue1Brown, ThePrimeagen.






























Who is shipping, who owns the news cycle, where the compute is, and what the money thinks happens next.
Live Polymarket odds on which company tops the leaderboard. Market prices, not predictions or advice.
Which company has the best AI model end of November? · $1.7M traded
Which company has the best AI model on December 15? · $43k traded
Share of today’s stories mentioning each lab. Merged coverage counts once.
Where the 50 hottest Hugging Face models come from. Fine-tunes and quantized builds count for the base model’s lab.
Largest known training run each year, in FLOP (log scale). From Epoch AI’s 1.1k notable models.
| Year | US | China | Europe |
|---|---|---|---|
| 2016 | 1e21.82 | 1e19.81 | 1e21.28 |
| 2017 | 1e20.93 | – | 1e20.81 |
| 2018 | 1e21.94 | – | 1e20.23 |
| 2019 | 1e22.83 | 1e19.58 | 1e23.03 |
| 2020 | 1e23.5 | 1e21.22 | 1e21.48 |
| 2021 | 1e24.31 | 1e24.02 | 1e23.8 |
| 2022 | 1e24.44 | 1e23.55 | 1e23.76 |
| 2023 | 1e25.7 | 1e24.11 | 1e23.89 |
| 2024 | 1e25.58 | 1e25.4 | 1e22.7 |
| 2025 | 1e26.7 | 1e25.18 | – |
| 2026 | 1e26.97 | 1e25.3 | 1e25.09 |
Latest additions to Epoch’s notable-model list
27 models landed on OpenRouter in the last 30 days, out of 470 listed. Price is dollars per million tokens, input / output.
| Added | Model | Maker | Context | $ / 1M tok |
|---|---|---|---|---|
| Nov 16 | Hy4 flashHy4 flash is a cost-efficient variant of [Hy4](https://openrouter.ai/tencent/hy4) from Tencent, served with roughly a quarter of the flagship's parameter count for lower latency and cost. It targets… | CN Tencent | 1M | $0.035 / $0.14 |
| Nov 15 | Qwen4 MaxQwen4 Max is the flagship model in Alibaba's Qwen4 series. It accepts text, image, and video input and returns text, with a 1M-token context window and reasoning enabled by default.... | CN Qwen | 1M | $1.425 / $4.275 |
| Nov 13 | Ling 3.2 Flash (free)Ling 3.2 Flash is a hybrid reasoning model from inclusionAI that accepts text and image input, with a 1M-token context window and up to 200 images per request. Thinking can... | CN inclusionAI | 1M | free |
| Nov 12 | Gemini 4 HydrogenGemini 4 Hydrogen is the high-efficiency workhorse model of the Gemini 4 family from Google, built for high-throughput agentic workflows, coding and multi-model routing at lower cost and latency… | US Google | 1M | $1.5 / $7.5 |
| Nov 10 | ERNIE-5.5-Flash-Preview (free)ERNIE 5.5 Flash Preview is the latest foundation model in the ERNIE family, with comprehensive improvements in coding, agentic capabilities, knowledge work, and deep research. It has 280 billion… | CN Baidu | 1M | free |
| Nov 10 | ERNIE-Light-DT-1.0ERNIE-Light-DT-1.0 is based on open-source GLM models and includes research findings from Baidu Search's Lightning large model. It features Baidu Sparse Attention (BSA), a unique sparse attention… | CN Baidu | 1M | $1.13 / $3.94 |
| Nov 10 | Gemini 4 CarbonGemini 4 Carbon is a frontier Gemini 4 model from Google for coding, agentic and complex reasoning work. It takes text, image, audio, video and file input, supports long decode... | US Google | 1M | $2 / $10 |
| Nov 9 | Nex-N2.6Nex-N2.6 is an agentic model from Nex AGI, built on [GLM-5.3-Flash](https://huggingface.co/zai-org/GLM-5.3-Flash-BF16) with about 321B parameters. It is suited for computer use, coding, and… | OTH Nex AGI | 262k | $0.12 / $0.4 |
| Nov 9 | Inkling 2 SmallInkling 2 Small is an open-weight multimodal mixture-of-experts reasoning model from Thinking Machines Lab, with 17B active parameters out of 594B total. It is the first model in the second... | OTH Thinking Machines | 1M | $0.3 / $1 |
| Nov 9 | Claude Fable 5.2Claude Fable 5.2 is Anthropic's most capable model for the most demanding technical work, with gains over Fable 5.1 in coding, reasoning, and vision. In prerelease testing it completed… | US Anthropic | 1M | $8 / $40 |
Notable models released in the last 12 months: 125.
The busiest AI prediction markets on Polymarket and their current front-runner. Market prices, not predictions or advice.
Everything in today’s issue. Filter it, search it, and share the link: the URL keeps your filters.
Showing all 783 items; the 25 already shown above aren’t repeated.
How OpenAI is working with NATO and allied cyber agencies to help trusted defenders find vulnerabilities, test fixes and protect essential systems.
text-generation · license apache-2.0

This episode is sponsored by Deepgram. Try Deepgram for voice agents that truly understand your context! #contextiseverything Not many people know that Stable Diffusion is why I went fulltime into AI (I aptly titled Multiverse, not Metaverse to call out…
AI assistant and MCP server for Blender: visual chat, scene tools, multi-agent workflows. Works with ChatGPT/GPT, Claude, Gemini, Ollama, OpenCode. Blender AI addon, copilot for 3D, bpy Python scripting, MCP integration.
Introducing computer-use agents, improved clinical answers, and new healthcare plugins that work across healthcare software to get real work done.
A 132-page export control rule landed Monday morning, and the Commerce Department’s own press release credited Treasury and the NSC with writing it. Days after the Megaspeed scandal ended Jensen Huang’s run at Nvidia, Washington is moving to close the cloud…

It doesn't even have to be 16 megapixels, as if 8 were enough.

At SC26, an annual supercomputing conference taking place this week in Chicago, NVIDIA is introducing the DGX Spark HPC Developer Kit for researchers, academics and developers. The kit is a four-node cluster of NVIDIA DGX Spark systems , each connected…
image-text-to-text · license apache-2.0
A little copying is better than a little dependency. — Go Proverbs , via Tags: go , programming

Having struggled for months reviewing PRs with thousands of lines of code that made no sense, delegated reviews to AI, and more, I decided to build a tool that allowed me to focus on the changes that actually mattered,…

Deb Shoquist leads operations at NVIDIA, overseeing the supply chain and manufacturing behind its AI platforms. That description sums up the job, but it doesn’t begin to encompass what Shoquist has made possible over decades. Shoquist, her team and their…
Which coding-agent harness should you use for your model?
Release: llm 0.37.3 A bug fix release for LLM , mainly to cover a problem with the most recent Homebrew package of the software. Homebrew now ships the latest SQLite, 3.54.1 , which includes a fix that broke LLM's logging code . Short version: in SQLite…
The PCs ran AI detectors on the reviews and left this comment on undisclosed AI usage in reviews: AI use in this review An AI-text detector assesses this review as likely AI-generated, yet its AI-use documentation…
We’re expanding our ChatGPT for Academic Researchers program to offer 200,000 researchers worldwide free access to our frontier models.
Keep your coding agent up to date with the latest web best practices
Investors fret over tighter US chip export controls and potential slowdown in AI data centre build-out
Spectacles to help the visually impaired, plus four more smart-tech inventions

Plan holiday road trips, get route tips, find local appointments, and search your saved lists with new Ask Maps features.
Public repository for Agent Skills

To train an AI model, thousands of GPUs have to work together, requiring a network that’s fast, efficient and resilient. Lambda is tackling the network piece of the puzzle with NVIDIA Photonics co-packaged optics (CPO) switches. The cloud provider is using…

Quantum processor deployments at supercomputing centers have grown nearly sevenfold over the past five years, with nearly 30 supercomputing centers around the world now in the process of deploying over 50 quantum processors. Many of these centers use Slurm…

In its MLPerf benchmarks debut, with results released today, the NVIDIA Vera Rubin platform delivers leading performance across training, post-training and agentic inference. NVIDIA swept every MLPerf Training v6.1 benchmark and, in its first MLPerf…

The French CEO of Hugging Face, a mega-hub of open-source AI software, says the attack by OpenAI models earlier this year proved one thing: Humanity can’t let AI be controlled by the few.

Model ML, a startup that builds artificial intelligence agents to handle…
Explore how more than 1,800 scientists use ChatGPT in research. See findings on daily workflows, reported productivity, and expectations for discovery.

Kang Wei developed the business while his wife was involved…
Discover how Revolut uses OpenAI models and Codex to cut coding time by 30–60%, scale financial crime investigations, and innovate without compromising security.
You do not always want a bazooka. Sometimes, though, you absolutely want a bazooka. That’s what Claude Fable 5.2 is for. This is your bazooka. For most purposes, for a thousand household uses, Claude Fable 5.2 will not be appreciably better than Claude Opus…
I deploy CNN + Attention stack models on edge devices, to great effect. Taking advantage of convs and reducing the spatial dimensionality makes all the sense in the world to me. Having vanilla ViTs learning the…

Just inland from the Mississippi River, about 1,300 workers are racing to…
LLM 驱动的多市场股票智能分析系统:多源行情、实时新闻、决策看板与自动推送,支持零成本定时运行。 LLM-powered multi-market stock analysis system with multi-source market data, real-time news, decision dashboard, automated notifications, and cost-free scheduled runs.
generate a video from an image with a text prompt
Every source that fed this issue, and whether it answered.