AI News Today

Anthropic today released Claude Sonnet 5 , a new AI model that the company says delivers near-flagship performance at mid-tier prices — a move designed to give cost-conscious enterprise developers access to powerful agentic capabilities just as the San Francisco-based AI lab barrels toward an initial public offering that will test whether the private market's staggering AI valuations can survive public scrutiny. The release, which Anthropic describes as " the most agentic Sonnet model ye t," mak

Anthropic’s Claude Sonnet 5 brings stronger agentic capabilities, lower pricing, and improved safety, positioning the model as a cheaper alternative to Opus, GPT-5.5, and Gemini Pro.

Not every new model is all it's cracked up to be. Our tracker keeps each release in context with its peers, so you know which models are worth your time.

Article URL: https://www.anthropic.com/news/claude-sonnet-5 Comments URL: https://news.ycombinator.com/item?id=48736605 Points: 1057 # Comments: 613

Article URL: https://www.gilesthomas.com/2026/06/llm-from-scratch-34a-building-a-jax-training-loop-for-an-llm-training-run Comments URL: https://news.ycombinator.com/item?id=48736594 Points: 1 # Comments: 0

[In beta] The product is called Minima and is live at minima.sh to try. It's a general purpose harness which uses Mubit (mubit.ai) memory to manage model routing Comments URL: https://news.ycombinator.com/item?id=48736592 Points: 1 # Comments: 0

Article URL: https://pacslate.com Comments URL: https://news.ycombinator.com/item?id=48736591 Points: 1 # Comments: 0

Hello, please help my web browser project become better: https://nordstjernen.org/ Please give any advice for how to make this web browser project successful. I have been making this web browser for the past month, using Claude and Codex. Comments URL: https://news.ycombinator.com/item?id=48736520 Points: 3 # Comments: 1

The modern job search means creating a resume and cover letter that will appeal to AI.

Acti is betting the smartphone keyboard is the next home for AI assistants. The startup's new keyboard for iOS and Android works across apps and lets users create custom AI-powered shortcuts using natural language.

Anthropic recently published an excellent write-up on how they contain Claude Code and its sub-agents. One thing that stood out is that the architecture isn’t really about Claude—it describes a general pattern for securing autonomous agents: * Every agent gets its own identity. * Authority is delegated (and attenuated) to sub-agents instead of being inherited wholesale. * Tool access is authorized per action. * Every action is attributable back to the originating principal. We took that architec

<pre style='white-space:pre-wrap;width:81ex'>fix(gateway): reject deleted-agent exec followups before dedupe</pre>

<pre style='white-space:pre-wrap;width:81ex'>refactor(gateway): distill deleted-agent RPC guard</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(gateway): reject deleted-agent sessions before dedupe and attachment offload</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(gateway): guard deleted-agent sessions before agent reset commands</pre>

<pre style='white-space:pre-wrap;width:81ex'>test(gateway): model ops as configured in voice-wake agent fixture</pre>

<pre style='white-space:pre-wrap;width:81ex'>test(gateway): add live agent RPC deleted-agent integration proof</pre>

<pre style='white-space:pre-wrap;width:81ex'>fix(gateway): reject agent RPC runs for deleted-agent sessions Related #65524</pre>

TL;DR Miles is RadixArk’s open source framework for large-scale LLM RL post-training. It composes SGLang for rollout, NVIDIA Megatron-LM for training, Ray orchestration, and PyTorch-native extensibility behind a small, pluggable...

Article URL: https://techstackups.com/articles/useful-llm-prompts-for-editing-your-own-technical-writing/ Comments URL: https://news.ycombinator.com/item?id=48736393 Points: 2 # Comments: 0

New Microsoft research shows how attackers can hijack AI agents that act on a user's behalf, using nothing more than a poisoned tool description to make the agent quietly hand over company data to an outsider. The trick is that the agent never breaks a rule. Every step looks routine, so in a default setup no alarm may fire. The work comes from Microsoft Incident Response and its

According to a report by The Information, OpenAI has cut inference costs for its AI models by more than half. The company applied the optimizations to ChatGPT, where the number of Nvidia GPUs needed dropped to just a few hundred at times. The article OpenAI reportedly cut response costs for guest ChatGPT users by more than half appeared first on The Decoder .

<p><strong><a href="https://bambamramfan.github.io/ai-compass/">The AI Compass</a></strong></p> This political compass style quiz <a href="https://bambamramfan.tumblr.com/post/820505178072580096/the-ai-compass">by bambamramfan</a> is pretty neat - answer 29 questions about AI and AI ethics to see which of the 30 archetypes you best fit.</p> <p>I'm impressed that my answers on my first time through the quiz categorized me as "The Garage Tinkerer", patron saint myself!</p> <p><img src="https://sta

We just open-sourced the internal system we built at Assembled for running coding agents as a team. Coding agents worked well for individual engineers, but the surrounding workflow was a bit of a mess. We generally found that many engineers had different MCP connections and context for their agents, personal automations running that other people couldn’t access, and very little introspection for what a human’s input into the coding agent looked like. So we built an internal system that converted

<p> AI notebook platform for modern data science </p> <p> <a href="https://www.producthunt.com/products/clusy?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1184840?app_id=339">Link</a> </p>

GPU-accelerated query engines are often constrained by memory and I/O bandwidth. NVIDIA hardware advances—including high bandwidth memory (HBM), NVIDIA...

Article URL: https://www.typeui.sh/docs/guides/hermes Comments URL: https://news.ycombinator.com/item?id=48736203 Points: 1 # Comments: 0

Memory shortages due to the AI boom have brought about RAMageddon as prices are expected to soon rise even further.

Article URL: https://0xff.nu/hex/ Comments URL: https://news.ycombinator.com/item?id=48736055 Points: 3 # Comments: 0

Hey, I am working on building agent-qa https://github.com/vostride/agent-qa which is a self-improving QA agent for software teams. Processed over 500M+ tokens and thousands of test runs, would love to understand your thoughts on this and how you are approaching testing today. Comments URL: https://news.ycombinator.com/item?id=48736050 Points: 1 # Comments: 1

A peek into where software engineering is headed from inside the sector’s leading AI labs. Agents running in the cloud are a major trend, while coding harnesses are spreading beyond the craft

As a developer building projects with AI coding agents, I need to have the system running day and night. Do you think buying mac studio is worth it? Comments URL: https://news.ycombinator.com/item?id=48735969 Points: 3 # Comments: 3

A new teaser trailer confirmed that Wonka's The Golden Ticket will premiere on Netflix on September 23rd, following its Squid Game reality show in the trend of creating real competitions based on fictional torture scenarios. While the sets seen in the trailer are real and not some Glasgow-style AI fakes, the voiceover is AI-generated. Deadline […]

Google adds two new generative AI models. Nano Banana 2 Lite generates images in four seconds at $0.034 a pop. Gemini Omni Flash brings video generation and editing via text prompts to the API for the first time. Google recommends chaining both models to go from a quick image to an animated video. The article Google launches Nano Banana 2 Lite for fast AI images and Gemini Omni Flash for video via API appeared first on The Decoder .

I tried to make it look and feel similar to NotebookLM but inside ChatGPT Comments URL: https://news.ycombinator.com/item?id=48735911 Points: 1 # Comments: 0

CNET tests the Beatbot Sora 70 robot pool cleaner to see how well it cleans, how easy it is to use, and whether it's worth the investment.

Article URL: https://www.clusy.io/ Comments URL: https://news.ycombinator.com/item?id=48735795 Points: 4 # Comments: 0

Article URL: https://claude.com/product/claude-science Comments URL: https://news.ycombinator.com/item?id=48735770 Points: 551 # Comments: 167

<!-- SC_OFF --><div class="md"><p>Was reading some research papers put out by Anthropic (and some other organizations/researchers) and one thing I've noticed is that these research papers consistently all share the same quality:</p> <ul> <li>Oftentimes over 100 pages of pure words, interspersed with screenshots of very dense/hard to read prompts and replies. Extremely-dry writing style.</li> <li>Oftentimes almost zero math or even math symbol to be seen.</li> <li>Uses some proprietary model with

Life sciences has entered an era of computational scale, and for more than a decade, NVIDIA has built the full GPU-accelerated computing stack — spanning hardware, frameworks, libraries, models, microservices and domain-specific tools — to help researchers run more sophisticated workflows and iterate faster. This week, Anthropic announced Claude Science, an AI workbench for science […]