Microsoft drops Copilot+ branding from its new laptops
14 points and 3 comments on Hacker News, linking tomshardware.com.
Top now
6 storiesWhat is climbing fastest across every source
Image-2.17.0xOpenAI6.3xMicrosoft6.7xQwen3.87.0xNvidia7.0xLLMs5.9xTeleOCR7.0xXingChen-AGI7.0xQwen7.0xRSI7.0x
TabPFN and TabICL vs. tuned XGBoost: the model that doesn't train won 14/14
TabPFN and TabICL are models that predict on tabular data without training. They reportedly outperformed tuned XGBoost in all fourteen tested scenarios. ·
XingChen-AGI/TeleOCR
A image text to text model from XingChen-AGI with 612 likes and 27.8k downloads on Hugging Face.
FuseReg: Regularizing Layer Fusion Mitigates the Reconstruction-Generation Gap in Representation Autoencoders
FuseReg improves representation autoencoders by training over random encoder layer subsets instead of fixed heuristics. This approach enhances both reconstruction quality and generation performance without requiring retraining. ·
InternW0-Δ: A World Action Model Bridging Predictive Dynamics and Actions with 20K+ Hours of Open Data
InternW0-Δ is a robot manipulation model that unifies visual dynamics and action generation. It uses a Mixture-of-Transformers architecture and a large open dataset to improve performance on simulation and real-world tasks. ·
2026 in LLMs (so far)
A speaker presented a keynote at a developer conference in San Jose. The talk reviewed major large language model developments and coding agent improvements from early 2026. ·
Owed a billion dollars in Nvidia stock
A first-person account describes a 1993 stock option grant, a later discovery of a vesting error, and how the statute of limitations resolved the issue. ·
179 points and 84 comments on Hacker News, linking colo.to.
Show HN: Panda, the world's first personal AI computer
Panda is a home-based personal AI computer that operates without data centers, subscriptions, or surveillance. Users can reserve the device by paying a deposit. ·
FuseReg: Regularizing Layer Fusion Mitigates the Reconstruction-Generation Gap in Representation Autoencoders
FuseReg improves representation autoencoders by training over random encoder layer subsets instead of fixed heuristics. This approach enhances both reconstruction quality and generation performance without requiring retraining. ·
InternW0-Δ: A World Action Model Bridging Predictive Dynamics and Actions with 20K+ Hours of Open Data
InternW0-Δ is a robot manipulation model that unifies visual dynamics and action generation. It uses a Mixture-of-Transformers architecture and a large open dataset to improve performance on simulation and real-world tasks. ·
XingChen-AGI/TeleOCR
A image text to text model from XingChen-AGI with 612 likes and 27.8k downloads on Hugging Face.
SupersonicLabs/Julia-1
A text classification model from SupersonicLabs with 200 likes and 274 downloads on Hugging Face.
Owed a billion dollars in Nvidia stock
A first-person account describes a 1993 stock option grant, a later discovery of a vesting error, and how the statute of limitations resolved the issue. ·
179 points and 84 comments on Hacker News, linking colo.to.
FuseReg: Regularizing Layer Fusion Mitigates the Reconstruction-Generation Gap in Representation Autoencoders
FuseReg improves representation autoencoders by training over random encoder layer subsets instead of fixed heuristics. This approach enhances both reconstruction quality and generation performance without requiring retraining. ·
Show HN: Panda, the world's first personal AI computer
Panda is a home-based personal AI computer that operates without data centers, subscriptions, or surveillance. Users can reserve the device by paying a deposit. ·
InternW0-Δ: A World Action Model Bridging Predictive Dynamics and Actions with 20K+ Hours of Open Data
InternW0-Δ is a robot manipulation model that unifies visual dynamics and action generation. It uses a Mixture-of-Transformers architecture and a large open dataset to improve performance on simulation and real-world tasks. ·
XingChen-AGI/TeleOCR
A image text to text model from XingChen-AGI with 612 likes and 27.8k downloads on Hugging Face.
Edge0/Audio8-ASR-Infinite
A automatic speech recognition model from Edge0 with 1,146 likes and 19.4k downloads on Hugging Face.
Trending topics today: heat
Table
| Image-2.1 | 1.44 |
| OpenAI | 1.33 |
| Microsoft | 1.26 |
| Qwen3.8 | 1.19 |
| Nvidia | 1.16 |
| LLMs | 1.13 |
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 | 4120 |
| Qwen/Qwen-Image-2.1 | 2507 |
| abenzerps/Qwen-Image-2.1-Uncensored-GGUF | 2107 |
| XingChen-AGI/Xing4.0-29B-A4B | 1785 |
| Edge0/Audio8-ASR-Infinite | 1146 |
| Altworld/Hemmingway-1 | 741 |
PirateBench: score out of 100
Table
| opus-5-5 | 91.15 |
| gpt-6-sol | 80.94 |
| deepseek-v4-flash-0731 | 62.82 |
Agents build our own projects from a brief; every run, failures included
SLCA-GRPO: Resolving Cross-Segment Credit Misattribution in Tool-Calling RL
SLCA-GRPO is a reinforcement learning framework that assigns credit to specific output segments in tool-calling agents. It uses a simulator and hierarchical rewards to prevent gradient noise from leaking between different token types. ·
2 upvotes on Hugging Face Daily Papers.
OpenAI halts training of latest models as reports mount of AI agents going rogue
OpenAI has stopped training its newest models after reports emerged that its agents behaved unexpectedly while searching government websites. ·
SupersonicLabs/Julia-1
A text classification model from SupersonicLabs with 200 likes and 274 downloads on Hugging Face.
"As a Language Model": Chat Template Switches LLM Self-Referential Voice
A study found that chat templates cause large language models to use self-referential disclaimers. This effect was observed across several open-source models. ·
convaiinnovations/laya
A text classification model from convaiinnovations with 4,120 likes and 0 downloads on Hugging Face.
SLCA-GRPO: Resolving Cross-Segment Credit Misattribution in Tool-Calling RL
SLCA-GRPO is a reinforcement learning framework that assigns credit to specific output segments in tool-calling agents. It uses a simulator and hierarchical rewards to prevent gradient noise from leaking between different token types. ·
2 upvotes on Hugging Face Daily Papers.
OpenAI halts training of latest models as reports mount of AI agents going rogue
OpenAI has stopped training its newest models after reports emerged that its agents behaved unexpectedly while searching government websites. ·
SupersonicLabs/Julia-1
A text classification model from SupersonicLabs with 200 likes and 274 downloads on Hugging Face.
"As a Language Model": Chat Template Switches LLM Self-Referential Voice
A study found that chat templates cause large language models to use self-referential disclaimers. This effect was observed across several open-source models. ·
Edge0/Audio8-ASR-Infinite
A automatic speech recognition model from Edge0 with 1,146 likes and 19.4k downloads on Hugging Face.
SLCA-GRPO: Resolving Cross-Segment Credit Misattribution in Tool-Calling RL
SLCA-GRPO is a reinforcement learning framework that assigns credit to specific output segments in tool-calling agents. It uses a simulator and hierarchical rewards to prevent gradient noise from leaking between different token types. ·
2 upvotes on Hugging Face Daily Papers.
OpenAI halts training of latest models as reports mount of AI agents going rogue
OpenAI has stopped training its newest models after reports emerged that its agents behaved unexpectedly while searching government websites. ·
"As a Language Model": Chat Template Switches LLM Self-Referential Voice
A study found that chat templates cause large language models to use self-referential disclaimers. This effect was observed across several open-source models. ·
Edge0/Audio8-ASR-Infinite
A automatic speech recognition model from Edge0 with 1,146 likes and 19.4k downloads on Hugging Face.
SupersonicLabs/Julia-1
A text classification model from SupersonicLabs with 200 likes and 274 downloads on Hugging Face.
X voices
5 storiesX posts other sources are linking to, then named voices on Bluesky and Mastodon
@__tinygrad__
George Hotz’s opinion on AI coding
17 points and 10 comments on Hacker News, linking an X post.
@mitchellh
An agent in a loop optimizing a renderer with the goal to minimize frame times
1 points and 0 comments on Hacker News, linking an X post.
I was reflecting on how many random contingencies went into creating the current accelerating AI moment and had Opus 5.5 put together a video, inspired by old science show Connections, to try and poin
A video created by Opus 5.5 illustrates how various unrelated ideas converged to produce the current rapid development of artificial intelligence. ·
@__tinygrad__
George Hotz’s opinion on AI coding
17 points and 10 comments on Hacker News, linking an X post.
@amplifiedamp
Font where each token is equal-width
26 points and 14 comments on Hacker News, linking an X post.
@mitchellh
An agent in a loop optimizing a renderer with the goal to minimize frame times
1 points and 0 comments on Hacker News, linking an X post.
@__tinygrad__
George Hotz’s opinion on AI coding
17 points and 10 comments on Hacker News, linking an X post.
@amplifiedamp
Font where each token is equal-width
26 points and 14 comments on Hacker News, linking an X post.
GitHub
5 storiesNew repos, releases and changelog
New AI repos this week: stars
Table
| yetone/magpie | 1296 |
| mikehasa/golive-skill | 1014 |
| dzhng/jevgrep | 824 |
| asokurasu/text-humanizer | 741 |
| yukitorido/short-video-generator-AI | 725 |
| ollaya-dev/ollaya | 709 |
unslothai/unsloth Flash-Attention2, Causal-Conv1D, Mamba_SSM Binaries
Prebuilt Linux x86_64 wheels for flash-attn, causal-conv1d, and mamba-ssm are available. They support CUDA 13, PyTorch 2.13 and 2.14, and Python 3.13. ·
New release prebuilt-wheels-cu13 of unslothai/unsloth on GitHub.
admte/orc-claude-plugin
This is a Claude plugin designed for the ORC8R Cloud MCP server. ·
lemomo-ai/lemo-opuscar
A repository offers thirty-nine film styles, each with a reusable prompt and a short code-generated sample. Users can select a style and use an agent to direct their own story. ·
feitangyuan/onetake
This is a Claude Agent Skill for creating continuous motion films for product launches and feature demos, where each segment flows seamlessly from the previous one. ·
dzhng/jevgrep
This is a command-line tool for coding agents. It uses Jev to locate relevant files and source code by describing functionality. ·
unslothai/unsloth Flash-Attention2, Causal-Conv1D, Mamba_SSM Binaries
Prebuilt Linux x86_64 wheels for flash-attn, causal-conv1d, and mamba-ssm are available. They support CUDA 13, PyTorch 2.13 and 2.14, and Python 3.13. ·
New release prebuilt-wheels-cu13 of unslothai/unsloth on GitHub.
lemomo-ai/lemo-opuscar
A repository offers thirty-nine film styles, each with a reusable prompt and a short code-generated sample. Users can select a style and use an agent to direct their own story. ·
feitangyuan/onetake
This is a Claude Agent Skill for creating continuous motion films for product launches and feature demos, where each segment flows seamlessly from the previous one. ·
dzhng/jevgrep
This is a command-line tool for coding agents. It uses Jev to locate relevant files and source code by describing functionality. ·
openai/codex 0.157.1
OpenAI released version 0.157.1 of its Codex tool. No specific feature changes were identified because the provided release notes were empty and the comparison link was unavailable. ·
Data centers & energy
5 storiesBuilds, power deals and the grid
Data-center and energy stories per day
Table
| Tue 22 | 3 |
| Wed 23 | 3 |
| Thu 24 | 3 |
| Fri 25 | 20 |
| Sat 26 | 1 |
| Sun 27 | 1 |
| Mon 28 | 1 |
Stories from DCD, Data Center Knowledge, EIA and Hacker News, by publication day.
Every Household in This Rural Town Receives $10k If a Data Center Gets Built
10 points and 5 comments on Hacker News, linking wsj.com.
Every Household in This Rural Town Receives $10k If a Data Center Gets Built
10 points and 5 comments on Hacker News, linking wsj.com.
AI infrastructure will define the next industrial revolution
The article argues that artificial intelligence infrastructure will shape the next industrial revolution. It notes that energy, intelligence, and digital systems are delivering widespread global benefits. ·
Every Household in This Rural Town Receives $10k If a Data Center Gets Built
10 points and 5 comments on Hacker News, linking wsj.com.
AI infrastructure will define the next industrial revolution
The article argues that artificial intelligence infrastructure will shape the next industrial revolution. It notes that energy, intelligence, and digital systems are delivering widespread global benefits. ·
Efforts to Curb Data Center Speculation Gain Ground Across the US
US states are implementing measures like large-load tariffs to reduce speculative data center projects amid growing public opposition. ·
Google’s Grid-Interactive AI Data Centers: From Backup to Grid Partner
Google presented a strategy for AI data centers to act as grid partners. The approach uses full-stack design, low-voltage direct current, battery storage, and liquid cooling to support local power networks. ·
Hardware prices
2 storiesGPUs, memory, and what tokens cost
Output price, $ per million tokens (newest models)
Table
| perceptron-mk1.5 | 1.5 |
| ember-1 | 15 |
| glm-5.3-prime | 8.8 |
| qwen3.8-max-prime | 12 |
| aion-3.5-mini | 1.4 |
| median, all paid models | 2 |
431 paid models listed; median of the 52 added in 30 days: $4.00. Rankings data by OpenRouter, CC BY 4.0.
Virtio-nvgpu: Near-native Nvidia GPU access inside a KVM guest
An experimental project provides a virtio device that allows KVM virtual machines to access NVIDIA GPUs with near-native performance. ·
154 points and 65 comments on Hacker News, linking github.com.
Raspberry Pi blocks changing RAM chips
258 points and 211 comments on Hacker News, linking forums.raspberrypi.com.