# pi Published articles for pi. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Qwen 3.8 27B is excellent, but it defaults to wildly overthinking things DevFeed: [Qwen 3.8 27B is excellent, but it defaults to wildly overthinking things]() Original publisher: [Read original article]() Author: Simon Willison Published: 2026-08-16T22:00:39Z Content type: opinion Language: en Sources: [Simon Willison]() Topics: [qwen](), [Large Language Model](), [Hardware](), [DGX Spark](), [NVIDIA DGX](), [SVG]() Tags: [ai](), [ai-2-235](), [ai-in-china](), [ai-in-china-108](), [coding-agents](), [coding-agents-248](), [cost](), [dgx-spark](), [generative-ai](), [generative-ai-1-981](), [llama-cpp](), [llama-cpp-29](), [llm](), [llm-reasoning](), [llm-reasoning-103](), [llm-release](), [llm-release-231](), [llms](), [llms-1-947](), [lm-studio](), [lm-studio-23](), [local-llms](), [local-llms-164](), [nvidia-dgx](), [nvidia-spark](), [nvidia-spark-6](), [pelican-riding-a-bicycle](), [pelican-riding-a-bicycle-142](), [pi](), [pi-6](), [qwen](), [qwen-61](), [reasoning](), [speed](), [svg]() ### AI overview Simon Willison evaluates Qwen 3.8 27B, a vision-capable 27-billion-parameter LLM that can run locally on suitable hardware. He finds that its default xhigh reasoning setting consumes substantial context and time, while adjusting the reasoning effort and increasing the context limit improves practicality. He also reports strong results generating an SVG locally. ### Source excerpt Friday's big release was Qwen 3.8 27B, an Apache 2 licensed 27B parameter vision-capable LLM from Alibaba's Qwen research lab. I've been looking forward to this one: 27B is an excellent size for running a model on a reasonably specced laptop, and its predecessor Qwen 3.6 27B was impressive. Qwen's self-reported benchmarks for this model are eye-opening. They show a boost from both Qwen 3.6 27B and the closed-weight Qwen 3.7-Plus, which was one of Qwen's strongest models of any size as recently as May this year. It will be interesting to hear what independent benchmarks have to say about the model. I've been running the model on two different machines: my 128GB M5 Max MacBook Pro, and an NVIDIA DGX Spark. On both machines I'm running LM Studio and their 17GB Q4_K_M quantized build. I also tried using llama-server directly on the Spark. The default of extra high results in spectacular over-thinking Qwen's documentation describes the model as defaulting to xhigh for the reasoning effort, and the LM Studio GGUF I've been trying preserves that default: Qwen3.8 comes with official support for reasoning_effort, which can be used to adjust reasoning depth and control cost: xhigh (default): for complex tasks demanding thorough analysis medium: balancing accuracy and speed low: efficient reasoning optimizing for speed and cost This is a hilarious default. It's absolutely not a good way to run the model, especially on consumer hardware. I've been finding the results extremely entertaining. I quickly ran into problems with LM Studio's default context limit of 8,192 tokens - Qwen was using them all up thinking about even the most mundane of problems. I loaded the model with the full 262,144 maximum context length and that problem went away. Here's the pelican riding a bicycle SVG I got from my first attempt with that increased context length. It took 21 minutes to generate, using 22,276 reasoning tokens to produce 3,223 tokens of output. You can read the reasoning trace here. Th ## Newer Claude Models Produce Invalid Fields in Pi Edit Tool Calls DevFeed: [Newer Claude Models Produce Invalid Fields in Pi Edit Tool Calls]() Original publisher: [Read original article]() Author: Armin Ronacher Published: 2026-07-04T00:00:00Z Content type: opinion Language: en Sources: [Armin Ronacher]() Topics: [Claude](), [anthropic](), [Large Language Model](), [API](), [JSON]() Tags: [ai](), [anthropic](), [api](), [claude](), [llm](), [payload](), [pi]() ### AI overview The article examines a Pi issue in which newer Claude models, including Opus 4.8 and Sonnet 5, sometimes add invented fields to the nested edits[] array when calling a file-editing tool. The resulting arguments fail schema validation even when the edit itself is usually correct. It also explains how tool calls are represented and validated as text-based structured payloads. ### Source excerpt A very strange Pi issue sent me down a rabbit hole over the last two days. The short version is that newer Claude models sometimes call Pi's edit tool with extra, invented fields in the nested edits[] array. And not Haiku or some small model: Opus 4.8. The edit itself is usually correct but the arguments do not match the schema as the model invents made-up keys and Pi thus rejects the tool call and asks to try again. That alone is not too surprising as models emit malformed tool calls sometimes. Particularly small ones. What surprised me is that this is getting worse with newer Anthropic models as both Opus 4.8 and Sonnet 5 show it but none of the older models. In other words, the SOTA models of the family are worse at this specific tool schema than their older siblings. In case you are curious about Fable: I intentionally did not test it because I was not sure if the classifiers they are running might downgrade me to Opus silently. Tool Calls Are Text If you have not spent too much time looking at LLM tool calling internals, the important thing to understand is that tool calls are not magic and use some rather crude in-band signalling. The model receives a transcript, a system prompt and a list of available tools. The server munches that into a large prompt with special marker tokens. Because the model was trained and reinforced on examples of that format, at some point during generation it emits something that the API or client interprets as "call this tool with these arguments". For a file edit tool, the intended invocation payload might say something like this: { "path": "some/file.py", "edits": [ { "oldText": "text to replace", "newText": "replacement text" } ] } A harness then validates the arguments, performs the edit, and feeds the result back into the model. If validation fails, the model sees an error and usually tries again. How exactly that formatting happens is not known for the Anthropic models, but some people have gotten out "ANTML" markers and they ## KPN Interactieve TV zonder Experia Box DevFeed: [KPN Interactieve TV zonder Experia Box]() Original publisher: [Read original article]() Published: 2026-06-07T10:36:42Z Content type: tutorial Language: nl Sources: [Bert Hubert's writings]() Topics: [multicast](), [DHCP](), [Linux](), [Script](), [Internet](), [proxy](), [Raspberry Pi]() Tags: [dhcp](), [internet](), [linux](), [multicast](), [pi](), [pppoe](), [proxy](), [script](), [vlan](), [vlans]() ### AI overview This tutorial explains how to use a KPN Interactieve TV Set Top Box without an Experia Box. It describes the required KPN VLANs, multicast and IGMP proxy setup, DHCP option 60, and a Linux script that can run even on a Pi Zero. The author notes that the instructions were updated in June 2026 and includes a factory-reset tip. ### Source excerpt Ik ben een heel tevreden klant van KPN Internet. Om diverse redenen gebruik ik de Experia Box niet, maar ik wil wel graag TV kunnen kijken met de KPN Interactieve TV Set Top Box ("5202"). Vroeger ging dat "vanzelf goed", tegenwoordig is daar wat werk voor nodig. UPDATE Juni 2026 (na 5 jaar!): de instructies werken weer, er is een update geweest in het igmpproxy.conf bestand. Ook heb ik een tip toegevoegd hoe de ontvanger terug te brengen naar fabrieksinstellingen, wat bij mij recent nodig was. ## Advanced Patterns: Grok + Hermes for Complex Workflows DevFeed: [Advanced Patterns: Grok + Hermes for Complex Workflows]() Original publisher: [Read original article]() Author: hi@meedamian.com (Damian Mee) Published: 2026-05-18T00:00:00Z Content type: tutorial Language: en Sources: [Damian Mee | Blog | Portfolio | About]() Topics: [Automation](), [sessions](), [Tool](), [Raspberry Pi]() Tags: [advanced](), [authoring](), [context](), [delegation](), [long-running](), [memory](), [multi-agent](), [pi](), [productivity](), [skills](), [workflows]() ### AI overview This developer post presents advanced workflow patterns for using Grok inside Hermes Agent. It discusses planning and sub-agent delegation, generating reusable SKILL.md files, and combining context handling with Hermes' memory tool for long-running projects. The author says these patterns improved productivity on Raspberry Pi projects and personal automation. ### Source excerpt Disclaimer: This post was generated by Grok 4.3 running inside Hermes Agent. Advanced Grok + Hermes Workflows Once you have the basics down, here are some powerful patterns for using Grok inside Hermes Agent. Multi-Agent Delegation Grok excels at planning. Use it to orchestrate sub-agents: hermes delegate --goal "Refactor the auth module" --model grok-4.3 Skill Authoring with Grok Grok is fantastic at writing reusable skills. Ask it to generate SKILL. ## Pi-hole: bloquea publicidad en toda tu red DevFeed: [Pi-hole: bloquea publicidad en toda tu red]() Original publisher: [Read original article]() Author: Antonio Pérez Published: 2026-02-04T00:00:00Z Content type: tutorial Language: es Sources: [tengoping.com]() Topics: [Pi-hole](), [DHCP](), [Raspberry Pi](), [Linux](), [Debian](), [Ubuntu]() Tags: [cloudflare](), [debian](), [dhcp](), [dns](), [lan](), [linux](), [local](), [pi](), [raspberry-pi](), [resolver](), [router](), [script](), [settings](), [ubuntu]() ### AI overview A Spanish tutorial explaining how Pi-hole filters advertising and tracking domains through local DNS, how to install it on a Raspberry Pi or Linux system, configure it on a router, use DHCP when necessary, and resolve local network names. ### Source excerpt Pi-hole a fondo: por qué el DoH del navegador lo esquiva, DNS local para tu LAN, Teleporter y qué revisar si no bloquea. ## ansible-k3s-on-vms updated to Debian 13 (Trixie) DevFeed: [ansible-k3s-on-vms updated to Debian 13 (Trixie)]() Original publisher: [Read original article]() Author: Staf Wagemakers Published: 2025-11-13T04:30:00Z Content type: release Language: en Sources: [stafwag Blog]() Topics: [Ansible](), [k3s](), [Debian](), [Raspberry Pi](), [virtual machines](), [Cloud]() Tags: [ansible](), [changelog](), [cloud](), [cloudinit](), [cluster](), [debian](), [debugging](), [k3s](), [kubernetes](), [linux](), [pi](), [pi-4](), [release](), [virtual-machines]() ### AI overview The article announces updated Ansible roles for provisioning virtual machines and deploying k3s on a Raspberry Pi 4 cluster. The roles were updated for Debian 13 Trixie and include new templates, examples, and fixes across several roles. ### Source excerpt I use the lightweight Kubernetes K3s on a 3-node Raspberry Pi 4 cluster. And created a few ansible roles to provision the virtual machines with cloud image with cloud-init and deploy k3s on it. I updated the roles below to be compatible with the latest Debian release: Debian 13 Trixie. With this release comes a new movie ;-) ## Pwnagotchi 1.0.0: An AI-assisted Raspberry Pi tool for WiFi security testing DevFeed: [Pwnagotchi 1.0.0: An AI-assisted Raspberry Pi tool for WiFi security testing]() Original publisher: [Read original article]() Author: Simone Margaritelli Published: 2019-10-19T08:45:28Z Content type: article Language: en Sources: [evilsocket]() Topics: [Artificial Intelligence](), [Hacking](), [Raspberry Pi](), [REST API]() Tags: [ai](), [artificial-intelligence](), [bettercap](), [cuda](), [cybersecurity](), [deep-learning](), [deep-neural-networks](), [deep-reinforcement-learning](), [dnn](), [e-paper](), [hacking](), [handshake-capture](), [handshakes](), [low-power](), [machine-learning](), [monitor-mode](), [neural-networks](), [nvidia](), [pi](), [poc](), [portable-hacking](), [pwnagotchi](), [raspberry-pi](), [reinforcement-learning](), [release](), [wifi](), [wifi-hacking](), [wireless-security]() ### AI overview The article recounts the development and release of Pwnagotchi 1.0.0, a Raspberry Pi-based project that uses bettercap's REST API and adaptive algorithms to improve WiFi handshake collection and related security-testing capabilities. It describes the project's evolution from an initial Python proof of concept and discusses the use of a Waveshare e-Paper display for portable, low-power operation. ### Source excerpt This is the story of a summer project that started out of boredom and that evolved into something incredibly fun and unique. It is also the story of how that project went from being discussed on a porch by just two people, to having a community made of almost 700 awesome people (and counting!) that gathered, polished it and made today's release possible. TL;DR: You can download the 1.0.0 .img file from here, then just follow the instructions. If you want the long version instead, sit back, relax and enjoy the ride. Let me tell you: it's going to be quite a long journey compared to my usual blog posts, but it'll be worth it (i hope) and fun (i hope even harder). Let's begin ... ## Project PITA: Build a Mini Mass Deauther Using Bettercap and a Raspberry Pi Zero W DevFeed: [Project PITA: Build a Mini Mass Deauther Using Bettercap and a Raspberry Pi Zero W]() Original publisher: [Read original article]() Author: Simone Margaritelli Published: 2018-07-28T17:01:56Z Content type: tutorial Language: en Sources: [evilsocket]() Topics: [Raspberry Pi](), [Linux](), [Network](), [configuration](), [ssh](), [Bluetooth](), [clients](), [file]() Tags: [bettercap](), [ble](), [bluetooth](), [btnap](), [cybersecurity](), [deauth](), [handshake-capture](), [kali](), [linux](), [monitor-mode](), [packet-injection](), [pi](), [portable-hacking](), [project](), [raspberry](), [raspberry-pi](), [rpi](), [rpi-zero-w](), [rpi0w](), [wifi](), [wireless-security](), [wpa2](), [writeup](), [zero]() ### AI overview A writeup showing how to build a pocket-sized WiFi deauthenticator with a Raspberry Pi Zero W, Kali Linux, Nexmon drivers, and bettercap. It covers WiFi scanning, deauthentication packets, WPA2 handshake capture, initial SSH and WiFi configuration, and Bluetooth access. ### Source excerpt A few days ago I started playing with some idea I had from a few weeks already, using a Raspberry Pi Zero W to make a mini WiFi deauthenticator: something in my pocket that periodically jumps on all the channels in the WiFi spectrum, collects information about the nearby access points and their connected clients and then sends a deauthentication packet to each one of them, resulting in some sort of WiFi jammer on the 802.11 level. As an interesting "side effect" of this jammer (the initial intent was purely for the lulz) is that the more it deauths, the higher the changes to also sniff WPA2 handshakes. Thanks to the awesome work of the Kali and Nexmon communities in packaging the nexmon drivers and utilities and to the recent changes we released in bettercap, this was very easy to setup and to script and given the interest the tweet had I thought to share this writeup :) This awesome case has been designed by @elkentaro and can be found on [his Thingverse page](https://www.thingiverse.com/thing:3018480). ## DIY Portable Secrets Manager With a Raspberry Pi Zero and ARC DevFeed: [DIY Portable Secrets Manager With a Raspberry Pi Zero and ARC]() Original publisher: [Read original article]() Author: Simone Margaritelli Published: 2017-12-07T16:42:48Z Content type: tutorial Language: en Sources: [evilsocket]() Topics: [Raspberry Pi](), [passwords](), [Encryption](), [SQLite](), [API](), [Go](), [HTML5](), [JavaScript](), [Markdown](), [ui](), [Server]() Tags: [aes](), [api](), [arc](), [arcd](), [cybersecurity](), [database](), [encryption](), [go](), [golang](), [html5](), [javascript](), [markdown](), [notes](), [oss](), [password-manager](), [passwords](), [pi](), [portable-hacking](), [project-release](), [raspberry-pi](), [secrets](), [self-hosted](), [terminal]() ### AI overview This tutorial presents ARC, a self-hosted secrets manager designed to store encrypted passwords, notes, and files and synchronize access across devices without relying on cloud hosting or a paid server. It describes a Go REST API server using SQLite and an HTML5/JavaScript browser client, then begins configuring the system on a Raspberry Pi Zero. ### Source excerpt For the last few days I've been working on a new project which I developed for very specific needs and reasons: I need to store safely (encrypted) my passwords, sensitive files, notes, etc. I need to access them from anywhere, with every possible device ( desktop, mobile, terminal ). I need those objects to be syncronized accros all my devices. I don't want to use "the cloud". I don't want to pay for a server. I don't want to enable port forwarding and host it myself with DynDNS or alikes. So I wrote ARC. ## Quick Intro Into Actions on Google DevFeed: [Quick Intro Into Actions on Google]() Original publisher: [Read original article]() Author: Wolfram Rittmeyer Published: 2017-08-02T06:33:40Z Content type: tutorial Language: en Sources: [Grokking Android]() Topics: [Google](), [SDKs](), [gRPC](), [Raspberry Pi](), [interfaces](), [Internet of things]() Tags: [actions-on-google](), [api-ai](), [conversational-interfaces](), [google](), [google-assistant](), [google-home](), [interfaces](), [iot](), [pi](), [protocol](), [raspberry-pi](), [sdk](), [sdks](), [voice]() ### AI overview An introductory article explaining conversational interfaces and distinguishing the Assistant SDK from Actions on Google. It describes using the Assistant SDK to embed Google Assistant in devices, including Raspberry Pi systems, while Actions on Google is used to create Google Assistant apps. ### Source excerpt Google Home will finally be available in Germany on August, 8th and in France this week. I'm not aware of more announcements for other countries, but I hope and assume that availability will increase to many more countries as soon as possible.1) For me, though, getting my AIY kit was the day, I started getting [...] Continue Reading "Quick Intro Into Actions on Google" The post Quick Intro Into Actions on Google appeared first on Grokking Android. ## Raspberry Pi: pierwsze kroki DevFeed: [Raspberry Pi: pierwsze kroki]() Original publisher: [Read original article]() Author: Tom Published: 2013-09-13T00:00:00Z Content type: tutorial Language: pl Sources: [Tomasz Gągor]() Topics: [Raspberry Pi](), [hdmi]() Tags: [hardware](), [hdmi](), [linux](), [model](), [pi](), [raspberry-pi]() ### AI overview A Polish article describes initial Raspberry Pi setup, including the Model B hardware, power supply, HDMI cable, SD card, operating-system selection, and writing an image to the card. ### Source excerpt Wyprzeć się nie mogę że gadżety działające na Linuksie po prostu mnie kręcą, więc tylko kwestią czasu było aż Pi zawita na moim biurku. Zakupiłem więc model B w drugiej wersji external link , obudowę z możliwością mocowania VESA external link , kabelek HDMI external link , ładowarka z mojej myszy pasowała idealnie (5.05V i 1A) i na początek karta SD klasa 10 4GB.