Serving GLM-5.3 at 607.8 tok/s across two Nvidia DGX stations with vLLM
vLLM serves GLM-5.3 at 607.8 tokens per second across two Nvidia DGX stations using ConnectX-8 RDMA. ·
27 points and 7 comments on Hacker News, linking developer.nvidia.com.
The front page of the internet for AI users · Sat 14 Nov 2026, 15:00 UTC · Get it by email
What is climbing fastest across every source
vLLM serves GLM-5.3 at 607.8 tokens per second across two Nvidia DGX stations using ConnectX-8 RDMA. ·
27 points and 7 comments on Hacker News, linking developer.nvidia.com.
Agweb is a browser designed for agents that avoids focus stealing, maintains a tiny footprint, and displays all tabs in a single view. ·
From Interconnects.
Hugging Face hosts the Haru-4B audio-to-audio model, which contains 5B parameters. ·
The Claude Delusion argues that its creator believes AI is amazing because he built it, yet also considers it superhuman because it surprised him. ·
91 points and 66 comments on Hacker News, linking netmeister.org.
vLLM serves GLM-5.3 at 607.8 tokens per second across two Nvidia DGX stations using ConnectX-8 RDMA. ·
A new repository written in Python with 389 stars in its first week.
Leonardo AI 2026 is a Windows 11 desktop application featuring text-to-image generation, a canvas editor, and real-time image-to-motion capabilities. ·
ChatRLabs releases Qwen-Image-2.1-Uncensored-Omni-Quantized, a text-to-image diffuser model with 7B parameters available on Hugging Face. ·
Anthropic releases Claude Code v2.1.325, which fixes a zstd security issue and adds a spinner tip for Jupyter notebooks. ·
An avalanche hits math, which now impacts AI development, leading to an expectation of superintelligence by 2030. ·
44 points and 45 comments on Hacker News, linking dwarkesh.com.
Nvidia allegedly engineered scarcity by funding AI startups to buy its GPUs. The Wall Street Journal reports this strategy inflated the ongoing GPU and memory shortage. ·
A new repository written in Python with 389 stars in its first week.
Leonardo AI 2026 is a Windows 11 desktop application featuring text-to-image generation, a canvas editor, and real-time image-to-motion capabilities. ·
Anthropic releases Claude Code v2.1.325, which fixes a zstd security issue and adds a spinner tip for Jupyter notebooks. ·
ggml-org releases llama.cpp version 1.0.0, which includes a nightly build labeled b12359 and a changelog since version 0.9.0. ·
| NVIDIA | 1.96 |
| Jensen | 1.17 |
| OpenAI | 1.01 |
| GPUs | 0.82 |
| XTwiNSaraX | 0.58 |
| buttheadtrenton16-byte | 0.54 |
Heat = summed hotness of today's stories naming the topic.
Releases, papers, trending weights and new listings
| Qwen/Qwen3.8-27B ▲2 | 19954 |
| Lightricks/LTX-2.5 ▲1 | 11870 |
| xtools-at/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic | 7261 |
| speridlabs/idol ▼1 | 2329 |
| MiniMaxAI/MiniMax-H3-Image | 1084 |
| NousResearch/Arete-27B ▼2 | 585 |
| 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/
The Claude Delusion argues that its creator believes AI is amazing because he built it, yet also considers it superhuman because it surprised him. ·
91 points and 66 comments on Hacker News, linking netmeister.org.
Anthropic upgrades Claude Code Projects to run parallel AI workers hands-free, with the coordinator now proactively answering questions and pushing thread status updates in chat. ·
ChatRLabs releases Qwen-Image-2.1-Uncensored-Omni-Quantized, a text-to-image diffuser model with 7B parameters available on Hugging Face. ·
Claude Meetings integrates Claude into video calls to take notes, answer questions, and build documents and whiteboards during team discussions. ·
Edge0 is a text generation model with 5B parameters available on Hugging Face. ·
The Gemini Decisions API uses the Gemini Decision model to provide fast, calibrated multimodal decisions. ·
568 points and 161 comments on Hacker News, linking blog.google.
The Claude Delusion argues that its creator believes AI is amazing because he built it, yet also considers it superhuman because it surprised him. ·
Claude's status page reports elevated errors on claude.ai, Claude Code, and platform.claude.com. ·
An author conducts 600 experiments to analyze JEPA world models, contrasting them with generative models that predict the future by generating pixels. ·
67 points and 61 comments on Hacker News, linking claude.com.
X posts other sources are linking to, then named voices on Bluesky and Mastodon
Neuron turns model into conspiracy theorist
2 points and 0 comments on Hacker News, linking an X post.
Show HN: 1:1 Webflow to Claude Code converter
2 points and 0 comments on Hacker News, linking an X post.
Evidence Chinese AI labs built their newest models on Nvidia's Blackwell chips
5 points and 0 comments on Hacker News, linking an X post.
Today we're launching Petrichor (YC F26), a wet lab for frontier AI labs
2 points and 1 comments on Hacker News, linking an X post.
Nvidia prepares to fire CEO Jensen Huang, WSJ reports
225 points and 4 comments on Hacker News, linking an X post.
Evidence Chinese AI labs built their newest models on Nvidia's Blackwell chips
5 points and 0 comments on Hacker News, linking an X post.
Neuron turns model into conspiracy theorist
2 points and 0 comments on Hacker News, linking an X post.
Show HN: 1:1 Webflow to Claude Code converter
2 points and 0 comments on Hacker News, linking an X post.
New repos, releases and changelog
| HybridRobotics/supermimic | 1482 |
| FrigadeHQ/yogurt | 1377 |
| agentes-chat/agentes-chat | 1373 |
| alexeyw/lot-map | 782 |
| Dodaios/ste80-humanizer | 509 |
| linknimiball1974/ai-texture-forge | 454 |
A new repository written in Python with 389 stars in its first week.
Leonardo AI 2026 is a Windows 11 desktop application featuring text-to-image generation, a canvas editor, and real-time image-to-motion capabilities. ·
Anthropic releases Claude Code v2.1.325, which fixes a zstd security issue and adds a spinner tip for Jupyter notebooks. ·
Doc-foundry uses adversarial multi-agent detection engineering to write and prove security rules. Blue authors Sigma rules while Red attacks them and a deterministic referee proves coverage. ·
OpenAI releases version 3.51.0 of its Python library, which expands agent, image, Live, and external key support. ·
A new repository written in Python with 389 stars in its first week.
Leonardo AI 2026 is a Windows 11 desktop application featuring text-to-image generation, a canvas editor, and real-time image-to-motion capabilities. ·
Anthropic releases Claude Code v2.1.325, which fixes a zstd security issue and adds a spinner tip for Jupyter notebooks. ·
ggml-org releases llama.cpp version 1.0.0, which includes a nightly build labeled b12359 and a changelog since version 0.9.0. ·
OpenAI Codex version 0.169.0 adds multiple side-chat tabs, custom model providers, and Cloud connection support to the terminal workspace. ·
Data centers, power, chips and what tokens cost
| Sun 08 | 2 |
| Mon 09 | 35 |
| Tue 10 | 32 |
| Wed 11 | 31 |
| Thu 12 | 29 |
| Fri 13 | 29 |
| Sat 14 | 6 |
Stories from DCD, Data Center Knowledge, EIA and Hacker News, by publication day.
| gemini-4-hydrogen | 7.5 |
| gemini-4-hydrogen:batch | 3.75 |
| ernie-light-dt-1.0-intl | 3.94 |
| gemini-4-carbon | 10 |
| gemini-4-carbon:batch | 5 |
| median, all paid models | 2.04 |
444 paid models listed; median of the 35 added in 30 days: $2.50. Rankings data by OpenRouter, CC BY 4.0.
Nvidia allegedly engineered scarcity by funding AI startups to buy its GPUs. The Wall Street Journal reports this strategy inflated the ongoing GPU and memory shortage. ·
824 points and 495 comments on Hacker News, linking techpowerup.com.
The European Union offers to build a data center in Moldova to protect citizens' personal data from cyberattacks. ·
Nvidia allegedly engineered scarcity by funding AI startups to buy its GPUs. The Wall Street Journal reports this strategy inflated the ongoing GPU and memory shortage. ·
824 points and 495 comments on Hacker News, linking techpowerup.com.
Forensic analysis of public AI models reveals that DeepSeek, Moonshot AI, MiniMax, and Ant Group use NVIDIA Blackwell GPUs to train their systems. ·
Data centers pay 38% of energy infrastructure costs in 2025, while rural and low-income households cover the rest. ·
586 points and 276 comments on Hacker News, linking michael.bacarella.com.
The European Union offers to build a data center in Moldova to protect citizens' personal data from cyberattacks. ·