1231
Speculative Decoding in vLLM on AMD GPUs
A practical guide to speculative decoding in vLLM on AMD GPUs, covering draft-and-verify mechanics, MTP, EAGLE-3, DFlash, DSpark, configuration, tuning, and ben
Top posts
Every blog post that took off on Hacker News, ranked by the votes of its best submission. 1,523 posts.
1232
The same AI request emits 8.5 times more CO2 in Texas than in Sweden
Swedish AI provider Berget AI releases an open-source method for calculating the carbon emissions of every individual AI request, developed with researchers from SEI, WattTime and Klimatkollen — and customers can now follow their emissions per request in the console.
1233
The Doom Justifies the Valuation
I’ve been in Berkeley for the last 2 weeks. I haven’t really been back here for a while, and it’s worse than you can believe. This is a cult of atheistic hedonists needing AI doom to be true to justify their life choices. Or acceleration. Or something. They need to make impact. I mean, narcissism of small differences…
1234
MIT's Ad Hoc Committee on AI Use in Teaching, Learning, and Research Training
Through five intense months of meetings, research, and outreach across the MIT community, the Ad Hoc Committee on AI Use in Teaching, Learning, and Research Training sought to understand the role of generative AI in the life and educational mission of the Institute and recommend how to navigate its challenges and…
1235
Benchmarking OpenTelemetry: Can AI trace your failed login?
A lot of vendors pitch AI SRE. We tested 14 models across 11 programming languages; even the best ones struggle with instrumenting code with the leading open-source standard, OpenTelemetry.
1236
Chain of Thought for Humans
These past few weeks, I’ve been conducting interviews, and I often open with “make sure you articulate your thoughts out loud, so I can follow along with your thinking”. This is primarily because I…
1237
LLMs predict my coffee
Why not benchmark with physical experiments?
1238
Starting from scratch: Training a 30M Topological Transformer
Tauformer is a topological transformer (see paper ) that replaces dot‑product attention with a Laplacian-derived scalar (taumode) per token/head, then attends using distances in that scalar space. Below is a post-style overview of the idea and the first training signals from a 30M-parameter run.
1239
Formal Verification Gates for AI Coding Loops
Formal verification gates for AI coding loops, in the language you're already shipping.
1240
The Agent SDK
The Agent SDK is the framework we use to build our own managed agents like Bugbot, Security Reviewer, and Rollouts. Define agents in code and deploy them with one command.
1241
OpenAI Frontier
OpenAI Frontier is an enterprise platform for building, deploying, and managing AI agents with shared context, onboarding, permissions, and governance.
1242
An autopsy of AI-generated 3D slop
Thinking of using AI for your Shopify 3D models? Read this first. We compare AI-generated 3D 'slop' vs. handcrafted models to show why the human touch is very much required to attain positive ROI.
1243
Hackers influence ChatGPT and Gemini to direct users to scam centers
Think you can trust AI to give you correct answer? Think again.
1244
Meta's Muse appears to use an OpenAI model labeled muse-special
While Muse was building my website, one subagent ran on a model called azure/muse-special. So I dug through the filesystem again.
1245
Ramp's Sheets AI Exfiltrates Financials
A vulnerability in Ramp's Sheets AI allowed the agent to insert formulas that made external network requests without user approval, creating the risk of data exfiltration via indirect prompt injection.
1246
Why are neural networks and cryptographic ciphers so similar? (2025)
Several neural network design principles are almost identical to cipher design principles.
1247
1248
OpenAI and Broadcom unveil LLM-optimized inference chip
OpenAI and Broadcom introduce Jalapeño, a custom AI chip built for LLM inference to improve performance, efficiency, and scale across AI systems.
1249
AI is too expensive
If you liked this piece, you should subscribe to my premium newsletter. It’s $70 a year, or $7 a month, and in return you get a weekly newsletter that’s usually anywhere from 5,000 to 18,000 words, including vast, detailed analyses of NVIDIA, Anthropic and OpenAI’s
1250
Securing our frontier model training environment
OpenAI explains how we strengthened containment for frontier model training and evaluation, including offline environments and high-compute red-teaming.
1251
The Competitive Moat That AI Can't Replicate
In a world racing to automate every interaction, the organisations that invest in genuine human connection are quietly building something no algorithm can copy.
1252
Claude Code: Anatomy of a Misfeature
On Canada Day (July 1), 2026, Anthropic shipped a surprising "easter egg" to users of Claude Code: `2.1.198` includes an efficiency bypass which allows agents to continue on without being blocked on direction from a human.
1253
Price per 1M tokens is meaningless
It stops being all about the vibes when the API bill hits you. Many companies are now discovering that AI can indeed be pricey. One habit that might be driving up your AI bill is comparing models by $X per 1M tokens ....
1254
1255
The AI Great Leap Forward
In 1958, Mao ordered every village to produce steel. The steel was useless. The crops rotted. Today's top-down AI mandates are producing the same pattern: ba...
1256
Build a Basic AI Agent from Scratch: Long Task Planning
In the previous part of the Build A Basic AI Agent From Scratch series, we added the essential tools to our agent to allow it to work…
1257
Fireworks AI security breach impacts Notion and their customers
A tool that connects everyday work into one space. It gives you and your teams AI tools—search, writing, note-taking—inside an all-in-one, flexible workspace.
1258
Making frontier cybersecurity capabilities available to defenders
Claude Code Security is one step towards our goal of more secure codebases and a higher security baseline across the industry.
1259
How to Train Your Robot
The enormous advance in AI capabilities over the last several years has in large part been driven by the availability of vast amounts of training data.
1260
Step 5 Preview: Advancing the Pareto Frontier
阶跃星辰于2023年4月成立，以“智能阶跃，十倍每个人的可能”为使命。阶跃星辰坚定自研超级模型，积极布局算力、数据等关键资源，发挥算法和人才优势，已完成 Step-1 千亿参数语言大模型和 Step-1V 千亿多模态大模型的研发，在图像理解、多轮指令跟随、数学能力、逻辑推理、文本创作等方面性能达到业界领先水平。