
Nous Research unveils Arete-27B, a post-trained Qwen3.8-27B model that reinforces Western values, saying the underlying Chinese model's bias layer is shallow
We post-trained over Alibaba's Qwen3.8-27B to reinforce Western values on a…
Issue #008 of Token Times: 754 stories from 125 feeds, ranked by heat. The lead: Nous Research unveils Arete-27B, a post-trained Qwen3.8-27B model that reinforces Western values, saying the underlying Chinese model's bias layer is shallow.
The stories with the most heat that day, plus one pick from every corner of the AI world.

We post-trained over Alibaba's Qwen3.8-27B to reinforce Western values on a…
image-to-3d · license apache-2.0
Models for physical AI tasks, such as robotics and autonomous driving, need not reason solely in language. They can use visual plans--points, bounding boxes, and trajectories--to represent spatial goals and anticipate motion. However, autoregressive vision-language models…

Command your Claude Code and Codex agents with gestures
A hands-on guide to training Mixture-of-Experts models

Chang Che is a writer and contributor to the New Yorker . In August, researchers at a California security firm used an OpenAI model to find a software bug in PHP, a programming language that undergird some seventy per cent of the world’s websites, including…
The 36 hottest stories after the front page, mixed across sections.
让你的命令行 Agent 在后台自己接活、自己跟进到底;平时不烦你,遇到拿不准的事才发卡片找你拍板。
Does anyone know what this NeurIPS Slides live is? I got an email saying "We are SlidesLive, the recording company partnered with NeurIPS to record your talk. We are ready to collect and edit your recording" with some…
image-text-to-text · license mit
datasette #3046: Don't serve link: rel=next HTTP headers to Firefox for HTML pages I was looking at the Firefox network panel today when I noticed that requests to a Datasette table page were triggering an HTTP prefetch to the next page of rows in that…

I'have been dreaming about using a Diffusion Model that will generates Weights instead of pixels. Diffusion models actually are vast used in order to generate from a natural language text description into images, and…
image-text-to-text · license other
I think I prefer IPS over OLED bc I don’t want to get burned pixels. 27 inch or 32 inch is fine unless it’s preferred to go bigger or should I just run a dual monitor set up. I also like to do some light gaming so…
Films made with code. Short films, long YouTube videos, ads, explainers and music videos: an AI coding agent writes the script, builds every frame in code, designs the sound and hands you an MP4.
text-generation · license apache-2.0

I built an infinite, self-evolving Backrooms game you can play in your browser: dimension9.space Opus 5.5 follows a haunting directive I wrote for it: keep building this world. New rooms, characters, entities and…
image-text-to-text · license other
image-text-to-text · license apache-2.0

I suspect that if this is real, it will just be the new Plus while the current Plus will be aggressively limited.
Their pay grade could be next
I am posting another update because my refund request remains unresolved and Cursor has still not provided a genuine human response. The charge was made in September, and I submitted my refund request on October 15.…
Libertarian seeks to turn vast land and energy resources into the foundation of his ‘unregulated’ AI hub
license apache-2.0
image-to-video · license other

Moving AI from text to 3D space, world models simulate real-world physics to teach robots and autonomous cars how to navigate reality.

Hi y’all. Thought Id share the progress on my current local VLA agent project! using agent and language somewhat loosely here. Technically it is an agent, and technically it does adhere to text prompts. However very…

Alibaba's new coding-agent benchmark, Co-Design Bench, tests AI agents on real user requests that already tripped them up in practice. Even GPT-6 Astra, the top performer, fully solves only 26.4 percent of tasks. On older benchmarks, the same models score…
license apache-2.0
Headless 3D and VFX engine for agents. Research preview.
Something I don't understand about a lot of AI doomer scenarios is how they seem to completely ignore basic economics, competition, and the sheer complexity of the global economy. Take the idea that a handful of…

Environmental groups worldwide are using artificial intelligence…
text-generation · license other
image-text-to-text · license apache-2.0
sure the usage limits are way too low for the price we pay. even on medium, it burns through usage way too quickly compared to the competition. but today, it is absolutely murdering it on extra high. anyone else…

GenRobot AI, a Beijing-based company providing data infrastructure for robot…
718 more stories, by section and heat. Merged coverage appears once, with the other outlets listed.
Each hub collects the latest stories on one subject.