AI companies leak data to advertisers [pdf]
69 points and 10 comments on Hacker News, linking jorgegarciaherrero.com.
The front page of the internet for AI users · Tue 29 Sep 2026, 11:00 UTC
What is climbing fastest across every source
69 points and 10 comments on Hacker News, linking jorgegarciaherrero.com.
OpenAI released version 0.159.0 of Codex, adding an opt-in feature that allows new input to steer the model during active responses or long-running code calls. ·
Utilities, hyperscalers, and developers are rethinking power delivery to support the 200 gigawatt scale of AI infrastructure. ·
Journalist Garrison Lovely discusses his book on the push to replace human labor with AGI. He argues that treating automation as inevitable risks society and democracy, advocating for targeted industrial policy and stronger social safety nets. ·
69 points and 10 comments on Hacker News, linking jorgegarciaherrero.com.
OpenAI released version 0.159.0 of Codex, adding an opt-in feature that allows new input to steer the model during active responses or long-running code calls. ·
191 points and 138 comments on Hacker News, linking anthropic.com.
Researchers introduced Active Taskless Distillation, a method transferring capabilities using single-word prompts. In coding tests, this approach yielded a 5.34 percentage point gain on HumanEval+. ·
YuE2 unifies symbolic and audio music generation by first writing a readable score before producing full-song audio. Experts preferred its output over versions without symbolic planning. ·
A website framework automatically identifies trending topics and generates daily reports. Users can customize its sources and selection criteria to create industry-specific news sites. ·
191 points and 138 comments on Hacker News, linking anthropic.com.
JoasASantos maintains a repository of AI models that are uncensored or fine-tuned for cybersecurity tasks. ·
OpenAI released version 0.159.0 of Codex, adding an opt-in feature that allows new input to steer the model during active responses or long-running code calls. ·
69 points and 10 comments on Hacker News, linking jorgegarciaherrero.com.
Researchers introduced Active Taskless Distillation, a method transferring capabilities using single-word prompts. In coding tests, this approach yielded a 5.34 percentage point gain on HumanEval+. ·
A website framework automatically identifies trending topics and generates daily reports. Users can customize its sources and selection criteria to create industry-specific news sites. ·
| OpenAI | 4.01 |
| World Labs | 1.79 |
| Learning | 1.26 |
| Anthropic | 1.07 |
| Claude Sonnet 5.5 | 1.06 |
| General Agents | 1.05 |
Heat = summed hotness of today's stories naming the topic.
Releases, papers, trending weights and new listings
| convaiinnovations/laya | 4415 |
| Qwen/Qwen-Image-2.1 | 2624 |
| abenzerps/Qwen-Image-2.1-Uncensored-GGUF | 2338 |
| XingChen-AGI/Xing4.0-29B-A4B | 1803 |
| Edge0/Audio8-ASR-Infinite | 1458 |
| XingChen-AGI/TeleOCR | 830 |
| opus-5-5 | 91.15 |
| gpt-6-sol | 81.78 |
| deepseek-v4-flash-0731 | 62.82 |
Agents build our own projects from a brief; every run, failures included: https://cowerx.dev/daily/piratebench/
33 points and 31 comments on Hacker News, linking nytimes.com.
From Latent Space.
Researchers propose Physical Coding to represent robot task states as executable code. Their system, HexaAnything, uses this approach to call perception and control tools while revising actions based on external feedback. ·
Researchers introduced Active Taskless Distillation, a method transferring capabilities using single-word prompts. In coding tests, this approach yielded a 5.34 percentage point gain on HumanEval+. ·
Researchers introduced Domain-Normalized MOPD to fix unbalanced feedback in multi-teacher distillation. The method rescales domain-specific signals, improving average benchmark scores over standard MOPD across three model sizes. ·
191 points and 138 comments on Hacker News, linking anthropic.com.
Researchers introduced Active Taskless Distillation, a method transferring capabilities using single-word prompts. In coding tests, this approach yielded a 5.34 percentage point gain on HumanEval+. ·
YuE2 unifies symbolic and audio music generation by first writing a readable score before producing full-song audio. Experts preferred its output over versions without symbolic planning. ·
Researchers introduced Domain-Normalized MOPD to fix unbalanced feedback in multi-teacher distillation. The method rescales domain-specific signals, improving average benchmark scores over standard MOPD across three model sizes. ·
33 points and 31 comments on Hacker News, linking nytimes.com.
191 points and 138 comments on Hacker News, linking anthropic.com.
JoasASantos maintains a repository of AI models that are uncensored or fine-tuned for cybersecurity tasks. ·
Researchers introduced Active Taskless Distillation, a method transferring capabilities using single-word prompts. In coding tests, this approach yielded a 5.34 percentage point gain on HumanEval+. ·
YuE2 unifies symbolic and audio music generation by first writing a readable score before producing full-song audio. Experts preferred its output over versions without symbolic planning. ·
Researchers introduced Domain-Normalized MOPD to fix unbalanced feedback in multi-teacher distillation. The method rescales domain-specific signals, improving average benchmark scores over standard MOPD across three model sizes. ·
X posts other sources are linking to, then named voices on Bluesky and Mastodon
New Cyber-OSINT model released
9 points and 2 comments on Hacker News, linking an X post.
OpenAI: Tomorrow we are re-opening the Pro $200 subscription
52 points and 61 comments on Hacker News, linking an X post.
AMD Acquires Dr. Fei Fei Li's World Labs
11 points and 1 comments on Hacker News, linking an X post.
AMD is acquiring Fei-Fei Li's World Labs in an $8.2B, all-stock deal
2 points and 1 comments on Hacker News, linking an X post.
New Cyber-OSINT model released
9 points and 2 comments on Hacker News, linking an X post.
OpenAI: Tomorrow we are re-opening the Pro $200 subscription
52 points and 61 comments on Hacker News, linking an X post.
OpenAI models may get more reasoning budget through the API than through Codex
3 points and 0 comments on Hacker News, linking an X post.
AMD Acquires Dr. Fei Fei Li's World Labs
11 points and 1 comments on Hacker News, linking an X post.
Researchers proved AI has deleted every reason universities exist
2 points and 3 comments on Hacker News, linking an X post.
OpenAI: Tomorrow we are re-opening the Pro $200 subscription
52 points and 61 comments on Hacker News, linking an X post.
New Cyber-OSINT model released
9 points and 2 comments on Hacker News, linking an X post.
OpenAI models may get more reasoning budget through the API than through Codex
3 points and 0 comments on Hacker News, linking an X post.
AMD Acquires Dr. Fei Fei Li's World Labs
11 points and 1 comments on Hacker News, linking an X post.
Researchers proved AI has deleted every reason universities exist
2 points and 3 comments on Hacker News, linking an X post.
New repos, releases and changelog
| KKKKhazix/AIHOT | 2292 |
| yetone/magpie | 1830 |
| dzhng/jevgrep | 1588 |
| mikehasa/golive-skill | 1081 |
| ollaya-dev/ollaya | 927 |
| feitangyuan/onetake | 914 |
A website framework automatically identifies trending topics and generates daily reports. Users can customize its sources and selection criteria to create industry-specific news sites. ·
A new repository written in TypeScript with 2,292 stars in its first week.
A seven-node ESP32-S3 cluster runs a 0.4B language model using 1.58-bit ternary quantization over an SPI daisy-chain. ·
504 points and 192 comments on Hacker News, linking github.com.
LangChain released version 1.7.0 of its Fireworks integration. The update adds prompt caching middleware and fixes mid-stream read timeout classification. ·
Anthropic released Claude Code v2.1.284, making Claude Sonnet 5.5 the default model with a 1M token context window. ·
A website framework automatically identifies trending topics and generates daily reports. Users can customize its sources and selection criteria to create industry-specific news sites. ·
A new repository written in TypeScript with 2,292 stars in its first week.
504 points and 192 comments on Hacker News, linking github.com.
A seven-node ESP32-S3 cluster runs a 0.4B language model using 1.58-bit ternary quantization over an SPI daisy-chain. ·
Anthropic released Claude Code v2.1.284, making Claude Sonnet 5.5 the default model with a 1M token context window. ·
JoasASantos maintains a repository of AI models that are uncensored or fine-tuned for cybersecurity tasks. ·
JoasASantos maintains a repository of AI models that are uncensored or fine-tuned for cybersecurity tasks. ·
A new repository with 291 stars in its first week.
A website framework automatically identifies trending topics and generates daily reports. Users can customize its sources and selection criteria to create industry-specific news sites. ·
Anthropic released Claude Code v2.1.284, making Claude Sonnet 5.5 the default model with a 1M token context window. ·
504 points and 192 comments on Hacker News, linking github.com.
Unsloth added decision model support, a document library, and Apple Silicon optimizations. The update also enables local Laya execution and accelerates image and video generation by up to 4.5 times. ·
Builds, power deals and the grid
| Wed 23 | 3 |
| Thu 24 | 3 |
| Fri 25 | 20 |
| Sat 26 | 1 |
| Sun 27 | 1 |
| Mon 28 | 20 |
| Tue 29 | 5 |
Stories from DCD, Data Center Knowledge, EIA and Hacker News, by publication day.
From DCD.
Waterless two-phase liquid cooling supports high-density computing and sustained performance in low-latency trading environments. ·
AI infrastructure design should integrate power, cooling, and compute to optimize efficiency, resilience, and adaptability rather than focusing solely on capacity. ·
Goodman withdrew its application for a 90-megawatt data center campus in Sydney, citing significant changes to the local policy and regulatory environment. ·
From DCD.
Waterless two-phase liquid cooling supports high-density computing and sustained performance in low-latency trading environments. ·
AI infrastructure design should integrate power, cooling, and compute to optimize efficiency, resilience, and adaptability rather than focusing solely on capacity. ·
Goodman withdrew its application for a 90-megawatt data center campus in Sydney, citing significant changes to the local policy and regulatory environment. ·
10 points and 5 comments on Hacker News, linking wsj.com.
From DCD.
Waterless two-phase liquid cooling supports high-density computing and sustained performance in low-latency trading environments. ·
AI infrastructure design should integrate power, cooling, and compute to optimize efficiency, resilience, and adaptability rather than focusing solely on capacity. ·
GPUs, memory, and what tokens cost
| claude-sonnet-5.5 | 10 |
| claude-sonnet-5.5:batch | 5 |
| perceptron-mk1.5 | 1.5 |
| ember-1 | 15 |
| glm-5.3-prime | 8.8 |
| median, all paid models | 2 |
434 paid models listed; median of the 56 added in 30 days: $4.00. Rankings data by OpenRouter, CC BY 4.0.
Import AI issue 474 covers Platonic mindspace, space-based TPUs, and Zhipu's outer RSI loop. ·
From Import AI.
Import AI issue 474 covers Platonic mindspace, space-based TPUs, and Zhipu's outer RSI loop. ·
From Import AI.
Import AI issue 474 covers Platonic mindspace, space-based TPUs, and Zhipu's outer RSI loop. ·
From Import AI.
An experimental project provides a virtio device that allows KVM virtual machines to access NVIDIA GPUs with near-native performance. ·