Every SaaS business will become a harness around a model
86 points and 65 comments on Hacker News, linking blog.sshh.io.
Top now
6 storiesWhat is climbing fastest across every source
LLM6.4xOpenAI5.3xClaude6.8xCerebras6.6xTe Hiku Media7.0xGPT-66.2xMystery Hype Theater7.0xCTO7.0xIndigenous7.0xAnthropic5.3x
Three AI agents, two countries, and one uneven world wide web
The piece evaluates GPT, Claude, and Muse on multilingual research, source access, human oversight, and observability across two countries. ·
anthropics/claude-code v2.1.288
New release v2.1.288 of anthropics/claude-code on GitHub.
Stateless GitHub App installation tokens rolled out
GitHub completes the staged rollout of stateless GitHub App installation tokens, a process that began on April 27, 2026. ·
Copilot code review: API support and new default effort level
GitHub adds API support for Copilot code reviews and sets Balanced as the default effort level for these automated checks. ·
Open-sourcing AstaBrief, the fast report-generation model in Asta
Asta releases AstaBrief, an 8B open-weights model that generates cited scientific reports, available in its Fast mode or for independent infrastructure use. ·
17 points and 2 comments on Hacker News, linking allenai.org.
anthropics/claude-code v2.1.288
New release v2.1.288 of anthropics/claude-code on GitHub.
Every SaaS business will become a harness around a model
86 points and 65 comments on Hacker News, linking blog.sshh.io.
openai/openai-python v3.24.0
OpenAI releases version 3.24.0 of its Python library, adding custom voice creation and agent session events to the API. ·
Shardarcairn12/Discord-Server-Raider
Shardarcairn12 develops Discord-Server-Raider, a tool that uses AI algorithms to monitor Discord server data in real time. ·
Kutuyyy/Leaked-System-Prompt-AI
A repository collects leaked system prompts for ChatGPT, Claude, Gemini, Grok, Perplexity, Cursor, Lovable, and Replit to promote transparency. ·
Open-sourcing AstaBrief, the fast report-generation model in Asta
Asta releases AstaBrief, an 8B open-weights model that generates cited scientific reports, available in its Fast mode or for independent infrastructure use. ·
17 points and 2 comments on Hacker News, linking allenai.org.
anthropics/claude-code v2.1.288
New release v2.1.288 of anthropics/claude-code on GitHub.
Amazon seeks to offload $8B of Nvidia chips to investors
48 points and 39 comments on Hacker News, linking reuters.com.
Shardarcairn12/Discord-Server-Raider
Shardarcairn12 develops Discord-Server-Raider, a tool that uses AI algorithms to monitor Discord server data in real time. ·
Kutuyyy/Leaked-System-Prompt-AI
A repository collects leaked system prompts for ChatGPT, Claude, Gemini, Grok, Perplexity, Cursor, Lovable, and Replit to promote transparency. ·
Adaptive Reward Routing: Dynamic Multi-Reward Optimization for Joint Audio-Video Diffusion via Forward-Process RL
Researchers propose Adaptive Reward Routing, a method that dynamically adjusts update locations and reward coordination during the training of joint audio-video diffusion models. ·
Trending topics today: heat
Table
| LLM | 1.78 |
| OpenAI | 1.42 |
| Claude | 1.35 |
| Cerebras | 1.17 |
| Te Hiku Media | 1.17 |
| GPT-6 | 1.09 |
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
| Lightricks/LTX-2.5 ▲1 | 5993 |
| convaiinnovations/laya | 5005 |
| abenzerps/Qwen-Image-2.1-Uncensored-GGUF | 2841 |
| Qwen/Qwen-Image-2.1 ▼2 | 2837 |
| Cloudflare/clef ▲7 | 781 |
| Contrastive-LM/CLM-v0.1-8B ▲1 | 668 |
PirateBench: score out of 100
Table
| 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/
Open-sourcing AstaBrief, the fast report-generation model in Asta
Asta releases AstaBrief, an 8B open-weights model that generates cited scientific reports, available in its Fast mode or for independent infrastructure use. ·
17 points and 2 comments on Hacker News, linking allenai.org.
GPT-6 Astra plays World of Warcraft for the first time with agent-wow
GPT-6 Astra clears the Orc starting zone in World of Warcraft in 40 minutes using agent-wow. ·
Decision models like Jev don't beat LLM-as-a-judge or traditional classifiers
14 points and 1 comments on Hacker News, linking developers.redhat.com.
DeepSeek Harness Desktop for macOS and Windows
DeepSeek Harness is a desktop application for macOS and Windows that processes documents, spreadsheets, and code while scheduling tasks through composable plugins. ·
Cloudflare/clef-flash
A image text to text model from Cloudflare with 259 likes and 1,303 downloads on Hugging Face.
Open-sourcing AstaBrief, the fast report-generation model in Asta
Asta releases AstaBrief, an 8B open-weights model that generates cited scientific reports, available in its Fast mode or for independent infrastructure use. ·
17 points and 2 comments on Hacker News, linking allenai.org.
GPT-6 Astra plays World of Warcraft for the first time with agent-wow
GPT-6 Astra clears the Orc starting zone in World of Warcraft in 40 minutes using agent-wow. ·
Decision models like Jev don't beat LLM-as-a-judge or traditional classifiers
14 points and 1 comments on Hacker News, linking developers.redhat.com.
Adaptive Reward Routing: Dynamic Multi-Reward Optimization for Joint Audio-Video Diffusion via Forward-Process RL
Researchers propose Adaptive Reward Routing, a method that dynamically adjusts update locations and reward coordination during the training of joint audio-video diffusion models. ·
On-Policy or Off-Policy Learning? A Systematic Study of Distillation Dynamics
Researchers find that token-level KL direction shapes performance in Llama3 and Qwen2.5 distillation, while rollout policy plays a less central role. ·
X voices
5 storiesX posts other sources are linking to, then named voices on Bluesky and Mastodon
@deamosV
Show HN: Client rooms where calls and AI agents suggest edits, people approve
2 points and 2 comments on Hacker News, linking an X post.
@arcprize
Over the course of 19 months, the cost to reach 75% on ARC-AGI-1 fell 99.95%
2 points and 0 comments on Hacker News, linking an X post.
@GayaniFigma
Figma restricts MCP access to whitelisted clients, excluding Pi
186 points and 104 comments on Hacker News, linking an X post.
@garymarcus
Anthropic's proposed valuation is easily calculated
1 points and 0 comments on Hacker News, linking an X post.
@deamosV
Show HN: Client rooms where calls and AI agents suggest edits, people approve
2 points and 2 comments on Hacker News, linking an X post.
@GayaniFigma
Figma restricts MCP access to whitelisted clients, excluding Pi
186 points and 104 comments on Hacker News, linking an X post.
@arcprize
Over the course of 19 months, the cost to reach 75% on ARC-AGI-1 fell 99.95%
2 points and 0 comments on Hacker News, linking an X post.
GitHub
5 storiesNew repos, releases and changelog
New AI repos this week: stars
Table
| KKKKhazix/AIHOT | 4959 |
| Louis-CFM/coucou | 2973 |
| feder-cr/dots | 2478 |
| rehan-remade/universal-modder | 2126 |
| CopilotKit/OpenDots | 1457 |
| firelex/jeff | 1321 |
Show HN: Made an open-source Lego AI generator
A developer releases an open-source tool that generates LEGO models using Astra and Opus 5.5, powered by Jev. ·
65 points and 37 comments on Hacker News, linking github.com.
openai/openai-python v3.24.0
OpenAI releases version 3.24.0 of its Python library, adding custom voice creation and agent session events to the API. ·
Shardarcairn12/Discord-Server-Raider
Shardarcairn12 develops Discord-Server-Raider, a tool that uses AI algorithms to monitor Discord server data in real time. ·
Kutuyyy/Leaked-System-Prompt-AI
A repository collects leaked system prompts for ChatGPT, Claude, Gemini, Grok, Perplexity, Cursor, Lovable, and Replit to promote transparency. ·
openai/codex 0.160.0
OpenAI releases Codex version 0.160.0, which adds keyboard-accessible browsing for older tasks and middle-click pasting in fullscreen mode on Linux terminals. ·
Shardarcairn12/Discord-Server-Raider
Shardarcairn12 develops Discord-Server-Raider, a tool that uses AI algorithms to monitor Discord server data in real time. ·
A new repository with 227 stars in its first week.
Show HN: Made an open-source Lego AI generator
A developer releases an open-source tool that generates LEGO models using Astra and Opus 5.5, powered by Jev. ·
Kutuyyy/Leaked-System-Prompt-AI
A repository collects leaked system prompts for ChatGPT, Claude, Gemini, Grok, Perplexity, Cursor, Lovable, and Replit to promote transparency. ·
openai/openai-python v3.24.0
OpenAI releases version 3.24.0 of its Python library, adding custom voice creation and agent session events to the API. ·
anthropics/claude-code v2.1.287
New release v2.1.287 of anthropics/claude-code on GitHub.
Infrastructure
5 storiesData centers, power, chips and what tokens cost
Data-center and energy stories per day
Table
| Sun 27 | 1 |
| Mon 28 | 20 |
| Tue 29 | 33 |
| Wed 30 | 34 |
| Thu 01 | 26 |
| Fri 02 | 32 |
| Sat 03 | 0 |
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
| pareto-26.10-preview | 3.2 |
| gpt-6.1-sol-pro | 10 |
| gpt-6.1-sol | 10 |
| claude-sonnet-5.5 | 10 |
| claude-sonnet-5.5:batch | 5 |
| median, all paid models | 2 |
437 paid models listed; median of the 52 added in 30 days: $5.00. Rankings data by OpenRouter, CC BY 4.0.
What if we stopped using GPUs? [video]
37 points and 8 comments on Hacker News, linking youtube.com.
Power approval set to delay Oracle's Wisconsin AI datacenter
40 points and 15 comments on Hacker News, linking theregister.com.
Poland’s National Data Processing Center receives EU funding grant
Poland's National Data Processing Center receives a 1.56 billion zloty grant from the European Funds for Digital Development program. ·
Power approval set to delay Oracle's Wisconsin AI datacenter
40 points and 15 comments on Hacker News, linking theregister.com.
What if we stopped using GPUs? [video]
37 points and 8 comments on Hacker News, linking youtube.com.
Meta Uses A.I. Data Centers to Avoid Billions in Federal Taxes
253 points and 244 comments on Hacker News, linking nytimes.com.
Most data centers refusing to say how much water, electricity they use
216 points and 199 comments on Hacker News, linking nltimes.nl.
SDF vs. MSDF vs. Slug: GPU Text Rendering
The piece compares texture-atlas, SDF, MSDF, and Slug glyph rendering methods on the GPU, explaining how each works and when to use them. ·
