← 2026-03-06

2026-03-07

20 trending topics

2026-03-08 →
🤖
236 📈 rising vol: 12 | src: 3

Tell HN: I'm 60 years old. Claude Code has ignited a passion again

Also: Claude Code just shipped /loop - schedule recurring tasks for up to 3 days • Addicted to Claude Code–Help

24h: 1 3d: 4 7d: 25
hackernews reddit rss
🤖
138 📈 rising vol: 4 | src: 4

Is GPT-5.4 Worth It?

Also: Does ChatGPT 5.4 Support Interruptions? • Chatgpt 5.4 Thinking Extended - Mona Lisa ASCII

24h: 3 3d: 12 7d: 145
gmail reddit rss youtube
🤖
136 📈 rising vol: 5 | src: 3

If people just read the model prompting guide from OpenAI, over 95% of output complaints in here would disappear

Also: OpenAI introduces newGPT5.4 model to control computers - Zamin.uz • Caitlin Kalinowski: I resigned from OpenAI

24h: 6 3d: 30 7d: 178
gmail hackernews reddit
🤖
126 🆕 new vol: 6 | src: 2

Files are the interface humans and agents interact with

24h: 0 3d: 0 7d: 0
hackernews rss
🤖
63 🆕 new vol: 3 | src: 2

LLM Writing Tropes.md

24h: 0 3d: 0 7d: 0
hackernews rss
🤖
26 🆕 new vol: 1 | src: 2

Google PM open-sources Always On Memory Agent, ditching vector databases for LLM ...

24h: 0 3d: 0 7d: 0
gmail hackernews
🤖
23 📈 rising vol: 5 | src: 1

Anthropic: In evaluating Claude Opus 4.6 on BrowseComp, we found cases where the model recognized the test, then found and decrypted answers to it—raising questions about eval integrity in web-enabled environments.

Also: GPT-5.4 vs Claude Opus 4.6 vs Gemini 3.1 Pro: which model is actually best for each task? • Anthropic: In evaluating Claude Opus 4.6 on BrowseComp, we found cases where the model recognized the test, then found and decrypted answers to it—raising questions about eval integrity in web-enabled environments.

24h: 7 3d: 34 7d: 126
reddit
🤖
21 🆕 new vol: 1 | src: 2

Sarvam 105B, the first competitive Indian open source LLM

24h: 0 3d: 0 7d: 0
hackernews rss
🤖
21 🆕 new vol: 1 | src: 2

Verification debt: the hidden cost of AI-generated code

24h: 0 3d: 0 7d: 0
hackernews rss
🤖
21 🆕 new vol: 1 | src: 2

We're Training Students to Write Worse to Prove They're Not Robots

24h: 0 3d: 0 7d: 0
hackernews rss
🤖
21 🆕 new vol: 1 | src: 2

Why developers using AI are working longer hours

24h: 0 3d: 0 7d: 0
hackernews rss
🤖
20 📈 rising vol: 5 | src: 1

OpenAI Robotics head resigns after deal with Pentagon

Also: OpenAI head of robotics resigns after deal with Pentagon • OpenAI head of Hardware and Robotics resigns

24h: 6 3d: 30 7d: 178
reddit