The front page of the internet for AI users · Wed 14 Oct 2026, 01:00 UTC · Get it by email
What is climbing fastest across every source
Pulse Local is a Mac budgeting application that executes its artificial intelligence models directly on the user's device rather than sending queries to external servers. ·
29 points and 16 comments on Hacker News, linking pulseos.finance.
Physical Intelligence offers its latest robotics models through the Pi API beta, allowing users to upload data, fine-tune models, and deploy cloud-hosted inference on their own robots. ·
A text generation model from LiquidAI with 155 likes and 2,565 downloads on Hugging Face.
Now on OpenRouter at $2.5 in and $7.5 out per million tokens.
OpenRouter adds Pareto 26.9, a multimodal composite model designed for research, coding, and agentic workflows that delivers frontier-level performance on general tasks. ·
The llm-gemini plugin version 0.35 adds support for the new Gemini 4 Argon model, which costs $2 per million input tokens at launch. ·
Anthropic built Scion, a custom source control system, after daily code merges rose from 2,500 in May 2026 to over 30,000 by September. ·
510 points and 508 comments on Hacker News, linking anthropic.com.
190 points and 45 comments on Hacker News, linking developers.googleblog.com.
A text generation model from LiquidAI with 155 likes and 2,565 downloads on Hugging Face.
Unsloth releases a major update featuring up to ten times faster chat performance, improved browser automation, and support for new video generation models. ·
Egotoadcitadel offers a free, local tool that extracts named color palettes from images for designers and pixel artists without requiring cloud uploads. ·
A text classification model from autotrust with 300 likes and 20.5k downloads on Hugging Face.
Anthropic built Scion, a custom source control system, after daily code merges rose from 2,500 in May 2026 to over 30,000 by September. ·
510 points and 508 comments on Hacker News, linking anthropic.com.
190 points and 45 comments on Hacker News, linking developers.googleblog.com.
Unsloth releases a major update featuring up to ten times faster chat performance, improved browser automation, and support for new video generation models. ·
Egotoadcitadel offers a free, local tool that extracts named color palettes from images for designers and pixel artists without requiring cloud uploads. ·
A text generation model from LiquidAI with 155 likes and 2,565 downloads on Hugging Face.
A text classification model from autotrust with 300 likes and 20.5k downloads on Hugging Face.
| OpenAI | 3.14 |
| Gemini 4 Argon | 2.72 |
| Anthropic | 2.32 |
| OpenRouter | 1.39 |
| Claude | 1.34 |
| LFM2.5 | 1.21 |
Heat = summed hotness of today's stories naming the topic.
Releases, papers, trending weights and new listings
| Lightricks/LTX-2.5 ▲1 | 7645 |
| abenzerps/Qwen-Image-2.1-Uncensored-Quantized | 4204 |
| google/embeddinggemma-2 | 2105 |
| jialinyyzz/humanizer | 1516 |
| Qwen/Qwen-Image-2.1-Turbo | 709 |
| openbmb/MiniCPM-V-4.7-35B-A3B new | 462 |
| gpt-6.1-sol | 98.9 |
| opus-5-5 | 96.46 |
| gpt-6-sol | 91.6 |
| gpt-6-luna | 65.25 |
| haiku-4-5 | 4.28 |
Agents build our own projects from a brief; every run, failures included: https://cowerx.dev/daily/piratebench/
190 points and 45 comments on Hacker News, linking developers.googleblog.com.
961 points and 737 comments on Hacker News, linking deepmind.google.
40 points and 3 comments on Hacker News, linking parsimony.lasu.dev.
A text classification model from autotrust with 300 likes and 20.5k downloads on Hugging Face.
From Google DeepMind.
190 points and 45 comments on Hacker News, linking developers.googleblog.com.
961 points and 737 comments on Hacker News, linking deepmind.google.
A text classification model from autotrust with 300 likes and 20.5k downloads on Hugging Face.
From Google DeepMind.
A image text to text model from openbmb with 284 likes and 263 downloads on Hugging Face.
X posts other sources are linking to, then named voices on Bluesky and Mastodon
OpenAI finds recording device in one of their offices
82 points and 2 comments on Hacker News, linking an X post.
I just got our entire company's Claude Enterprise account suspended
10 points and 3 comments on Hacker News, linking an X post.
One plan now covers Cursor, Grok, Grok Bot, and X
2 points and 0 comments on Hacker News, linking an X post.
In 5 years, 95% of tokens used by AI agents will be for tasks humans never did
5 points and 0 comments on Hacker News, linking an X post.
OpenAI finds recording device in one of their offices
82 points and 2 comments on Hacker News, linking an X post.
I just got our entire company's Claude Enterprise account suspended
10 points and 3 comments on Hacker News, linking an X post.
The gap in shared understanding of LLM capability is widening
715 points and 1,044 comments on Hacker News, linking an X post.
One plan now covers Cursor, Grok, Grok Bot, and X
2 points and 0 comments on Hacker News, linking an X post.
New repos, releases and changelog
| wheresryan22/anatomy | 8200 |
| mhtsec/ARTEX | 3871 |
| onehumanbeing/esp32-iphone-agent | 1520 |
| NVIDIA-Omniverse/usd-brep | 1143 |
| trhhubsd/ARTEXX | 1142 |
| rociiu/talorys | 959 |
Unsloth releases a major update featuring up to ten times faster chat performance, improved browser automation, and support for new video generation models. ·
New release v0.1.907-beta of unslothai/unsloth on GitHub.
Egotoadcitadel offers a free, local tool that extracts named color palettes from images for designers and pixel artists without requiring cloud uploads. ·
15 points and 4 comments on Hacker News, linking github.com.
DishwasherBench is a browser-based physical simulation game called Spotless that benchmarks language models by testing their ability to load a real dishwasher. ·
71 points and 30 comments on Hacker News, linking github.com.
Unsloth releases a major update featuring up to ten times faster chat performance, improved browser automation, and support for new video generation models. ·
New release v0.1.907-beta of unslothai/unsloth on GitHub.
Egotoadcitadel offers a free, local tool that extracts named color palettes from images for designers and pixel artists without requiring cloud uploads. ·
DishwasherBench is a browser-based physical simulation game called Spotless that benchmarks language models by testing their ability to load a real dishwasher. ·
15 points and 4 comments on Hacker News, linking github.com.
Aquafitzdisconnect releases Discord Server Raider, an AI-driven tool that automates server management and moderation tasks to improve user experience and resource efficiency. ·
Data centers, power, chips and what tokens cost
| Thu 08 | 28 |
| Fri 09 | 30 |
| Sat 10 | 3 |
| Sun 11 | 7 |
| Mon 12 | 41 |
| Tue 13 | 29 |
| Wed 14 | 0 |
Stories from DCD, Data Center Knowledge, EIA and Hacker News, by publication day.
One point per OpenRouter reading, last 8 days. Data by OpenRouter, CC BY 4.0.
| pareto-latest | 7.5 |
| pareto-26.9 | 7.5 |
| lfm-2.5-24b-a2b | 0.7 |
| gemini-4-argon | 10 |
| gemini-4-argon:batch | 5 |
| median, all paid models | 2 |
437 paid models listed; median of the 47 added in 30 days: $5.00. Rankings data by OpenRouter, CC BY 4.0.
Researchers describe a distributed virtual GPU cloud that utilizes idle Apple Silicon devices to address limitations in centralized machine learning inference infrastructure. ·
217 points and 92 comments on Hacker News, linking machinelearning.apple.com.
25 points and 39 comments on Hacker News, linking reuters.com.
Janas streams large language model experts from SSDs to run models exceeding available RAM, achieving 37 tokens per second on a 32 GB laptop. ·
Intel introduces the D270, a data center GPU designed for inference workloads, which supports up to 480 GB of LPDDR5X memory. ·
25 points and 39 comments on Hacker News, linking reuters.com.
Researchers describe a distributed virtual GPU cloud that utilizes idle Apple Silicon devices to address limitations in centralized machine learning inference infrastructure. ·
Small AI models may soon self-replicate across cheap IoT devices like Android TV boxes without requiring a datacenter. ·
Intel introduces the D270, a data center GPU designed for inference workloads, which supports up to 480 GB of LPDDR5X memory. ·
Janas streams large language model experts from SSDs to run models exceeding available RAM, achieving 37 tokens per second on a 32 GB laptop. ·