Top now
6 storiesWhat is climbing fastest across every source
Digital Realty Plans Cable Landing Station at Los Angeles Data Center
Digital Realty plans a turnkey cable landing station at its Los Angeles data center to connect subsea capacity directly to the interconnection environment. ·
How Carbon-Aware Scheduling Can Cut Data Center Emissions
Carbon-aware scheduling aligns flexible workloads to cleaner hours and locations, turning variability into measurable Scope 2 reductions for data centers. ·
[AINews] OpenAI DevDay 2026: Dots, 6.1 Sol, Ultrafast, Decisions API, Agents API, Spaces, Marketplace, and 1.2 Billion ChatGPT WAU
From Latent Space.
Responsible Release of AI-Generated Mathematics
Frontier AI labs are testing proprietary models on advanced mathematical problems. The text discusses the responsible release of AI-generated mathematics. ·
GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price
943 points and 828 comments on Hacker News, linking openai.com.
Jeeves. Reasoning improves Jev-like decision models
235 points and 91 comments on Hacker News, linking github.com.
ROSS: Relearning from Self-Generated Rollouts through Selective Supervision
Researchers introduced ROSS, a method that reuses historical model outputs during training by applying loss only to selected continuations. This approach improved the Qwen3.6-35B-A3B model's average benchmark score from 58.40% to 62.20%. ·
Raven: The Harness of Harnesses for Composable Agentic Intelligence
155 upvotes on Hugging Face Daily Papers.
feder-cr/dots
Feder-cr released an open-source AI agent that operates its own browser to avoid being blocked on the web. ·
openai/codex 0.159.2
OpenAI released version 0.159.2 of Codex, a patch that stops console windows from flashing on Windows when the tool runs background processes. ·
GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price
943 points and 828 comments on Hacker News, linking openai.com.
PostHog/jeeves
235 points and 91 comments on Hacker News, linking github.com.
Language models for text classification: From bag-of-words to Jev
The source text lists titles for visual guides and experiments on recurrent neural networks, convolutional neural networks, and transformers, focusing on their accuracy and efficiency. ·
ROSS: Relearning from Self-Generated Rollouts through Selective Supervision
Researchers introduced ROSS, a method that reuses historical model outputs during training by applying loss only to selected continuations. This approach improved the Qwen3.6-35B-A3B model's average benchmark score from 58.40% to 62.20%. ·
Raven: The Harness of Harnesses for Composable Agentic Intelligence
155 upvotes on Hugging Face Daily Papers.
JoasASantos/Offensive-Security-AI-Models
JoasASantos maintains a repository of AI models that are uncensored or fine-tuned for cybersecurity tasks. ·
Trending topics today: heat
Table
| OpenAI | 3.01 |
| Anthropic | 2.21 |
| GLM-5.3 | 1.26 |
| Video Generation | 1.15 |
| Language | 1.14 |
| LLMs | 1.14 |
Heat = summed hotness of today's stories naming the topic.
Models
5 storiesReleases, papers, trending weights and new listings
Top 6 trending on Hugging Face: likes
Table
| convaiinnovations/laya | 4565 |
| Qwen/Qwen-Image-2.1 ▼1 | 2685 |
| abenzerps/Qwen-Image-2.1-Uncensored-GGUF | 2482 |
| Edge0/Audio8-ASR-Infinite | 1561 |
| XingChen-AGI/TeleOCR ▲1 | 937 |
| Contrastive-LM/CLM-v0.1-8B ▲1 | 543 |
PirateBench: score out of 100
Table
| 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/
Jeeves. Reasoning improves Jev-like decision models
235 points and 91 comments on Hacker News, linking github.com.
PSSA: A non-transformer language model written from scratch in Rust
76 points and 32 comments on Hacker News, linking github.com.
ROSS: Relearning from Self-Generated Rollouts through Selective Supervision
Researchers introduced ROSS, a method that reuses historical model outputs during training by applying loss only to selected continuations. This approach improved the Qwen3.6-35B-A3B model's average benchmark score from 58.40% to 62.20%. ·
Raven: The Harness of Harnesses for Composable Agentic Intelligence
155 upvotes on Hugging Face Daily Papers.
MaLiang-Harness: A Programmable Path to Image and Video Generation
Researchers introduced MaLiang-Harness, a framework that uses multimodal models to iteratively build, check, and fix code for image and video generation. The system evaluates eleven closed-source models on new benchmarks to measure success and quality. ·
Language models for text classification: From bag-of-words to Jev
The source text lists titles for visual guides and experiments on recurrent neural networks, convolutional neural networks, and transformers, focusing on their accuracy and efficiency. ·
121 points and 5 comments on Hacker News, linking magazine.sebastianraschka.com.
ROSS: Relearning from Self-Generated Rollouts through Selective Supervision
Researchers introduced ROSS, a method that reuses historical model outputs during training by applying loss only to selected continuations. This approach improved the Qwen3.6-35B-A3B model's average benchmark score from 58.40% to 62.20%. ·
Claude Sonnet 5.5
191 points and 138 comments on Hacker News, linking anthropic.com.
Raven: The Harness of Harnesses for Composable Agentic Intelligence
155 upvotes on Hugging Face Daily Papers.
JoasASantos/Offensive-Security-AI-Models
JoasASantos maintains a repository of AI models that are uncensored or fine-tuned for cybersecurity tasks. ·
X voices
5 storiesX posts other sources are linking to, then named voices on Bluesky and Mastodon
Trump, Musk, Jensen. and Somewhere, Anthropic
2 points and 0 comments on Hacker News, linking an X post.
The Next 3x in Inference Won't Come from Faster Kernels
2 points and 1 comments on Hacker News, linking an X post.
New Cyber-OSINT model released
53 points and 16 comments on Hacker News, linking an X post.
Changing the difference between Codex plans to be Plus=1x Pro 100=5x Pro 200=10x
2 points and 2 comments on Hacker News, linking an X post.
OpenAI: Tomorrow we are re-opening the Pro $200 subscription
76 points and 91 comments on Hacker News, linking an X post.
OpenAI: Tomorrow we are re-opening the Pro $200 subscription
76 points and 91 comments on Hacker News, linking an X post.
New Cyber-OSINT model released
53 points and 16 comments on Hacker News, linking an X post.
Trump, Musk, Jensen. and Somewhere, Anthropic
2 points and 0 comments on Hacker News, linking an X post.
The Next 3x in Inference Won't Come from Faster Kernels
2 points and 1 comments on Hacker News, linking an X post.
Changing the difference between Codex plans to be Plus=1x Pro 100=5x Pro 200=10x
2 points and 2 comments on Hacker News, linking an X post.
GitHub
5 storiesNew repos, releases and changelog
New AI repos this week: stars
Table
| KKKKhazix/AIHOT | 3762 |
| dzhng/jevgrep | 1831 |
| feitangyuan/onetake | 1017 |
| kaankiziltug/logo-design-skill | 937 |
| feder-cr/dots | 734 |
| lemomo-ai/lemo-opuscar | 643 |
Jeeves. Reasoning improves Jev-like decision models
235 points and 91 comments on Hacker News, linking github.com.
PSSA: A non-transformer language model written from scratch in Rust
76 points and 32 comments on Hacker News, linking github.com.
feder-cr/dots
Feder-cr released an open-source AI agent that operates its own browser to avoid being blocked on the web. ·
openai/codex 0.159.2
OpenAI released version 0.159.2 of Codex, a patch that stops console windows from flashing on Windows when the tool runs background processes. ·
anthropics/claude-code v2.1.285
New release v2.1.285 of anthropics/claude-code on GitHub.
PostHog/jeeves
235 points and 91 comments on Hacker News, linking github.com.
JoasASantos/Offensive-Security-AI-Models
JoasASantos maintains a repository of AI models that are uncensored or fine-tuned for cybersecurity tasks. ·
feder-cr/dots
Feder-cr released an open-source AI agent that operates its own browser to avoid being blocked on the web. ·
openai/codex 0.159.2
OpenAI released version 0.159.2 of Codex, a patch that stops console windows from flashing on Windows when the tool runs background processes. ·
anthropics/claude-code v2.1.285
New release v2.1.285 of anthropics/claude-code on GitHub.
Infrastructure
5 storiesData centers, power, chips and what tokens cost
Data-center and energy stories per day
Table
| Thu 24 | 3 |
| Fri 25 | 20 |
| Sat 26 | 1 |
| Sun 27 | 1 |
| Mon 28 | 20 |
| Tue 29 | 33 |
| Wed 30 | 6 |
Stories from DCD, Data Center Knowledge, EIA and Hacker News, by publication day.
Median output price of models added in the last 30 days, $ per million tokens
One point per OpenRouter reading, last 8 days. Data by OpenRouter, CC BY 4.0.
Output price, $ per million tokens (newest models)
Table
| gpt-6.1-sol-pro | 10 |
| gpt-6.1-sol-pro:batch | 5 |
| gpt-6.1-sol | 10 |
| gpt-6.1-sol:batch | 5 |
| claude-sonnet-5.5 | 10 |
| median, all paid models | 2 |
438 paid models listed; median of the 60 added in 30 days: $5.00. Rankings data by OpenRouter, CC BY 4.0.
AI needs $6T in annual revenue to justify data centre boom
203 points and 300 comments on Hacker News, linking thenationalnews.com.
AI needs $6T in annual revenue to justify data centre boom
203 points and 300 comments on Hacker News, linking thenationalnews.com.