The best Hacker News stories from Show from the past week
Latest posts:
Show HN: Price Per Token – LLM API Pricing Data
The LLM providers are constantly adding new models and updating their API prices. Anyone building AI applications knows that these prices are very important to their bottom line. The only place I am aware of is going to these provider's individual website pages to check the price per token.<p>To solve this inconvenience I spent a few hours making pricepertoken.com which has the latest model's up-to-date prices all in one place.<p>Thinking about adding image models too especially since you have multiple options (fal, replicate) to use the same model and the prices are not always the same.
Show HN: TheProtector – Linux Bash script for the paranoid admin on a budget
Hi HN,<p>I spent the past year building this in my spare time because I got tired of enterprise security tools that cost $50K/year and don't understand Linux.<p>TheProtector is a comprehensive security monitoring tool that actually runs on the systems we use (Linux) instead of being a Windows-first afterthought. Built it entirely on a $500 laptop because I believe good security shouldn't require unlimited budgets.<p>Features:
- Real-time process, network, and file monitoring
- YARA malware detection with custom rules
- eBPF kernel monitoring (when available)
- Behavioral baseline establishment and anomaly detection
- Active threat response (blocks IPs, kills processes, quarantines files)
- Anti-evasion detection for rootkits and advanced threats
- Honeypots for attack detection
- Web dashboard for monitoring
- Single bash script, no complex installation<p>The tagline is "not perfect but better than most" because I'm tired of security vendors claiming their tools are flawless. This actually works, costs $0, and you can read every line of code.<p>I know bash isn't the sexy choice for security tools, but it runs everywhere, has zero dependencies, and most Linux admins can read/modify it. Sometimes boring technology that works is better than fancy technology that doesn't.<p>It's designed for the intersection of "paranoid about security" and "don't have enterprise budgets" - which describes most of us actually running Linux systems.<p>GitHub: <a href="https://github.com/IHATEGIVINGAUSERNAME/theProtector">https://github.com/IHATEGIVINGAUSERNAME/theProtector</a><p>Been running it on my own systems for months. Catches the stuff that matters and doesn't flood you with false positives. If you hate expensive security theater as much as I do, might be worth a look.<p>Open to feedback, especially from folks who know more about this stuff than I do.<p>Thanks,
IHATEGIVINGAUSERNAME (yes, I really do hate giving usernames)
Show HN: Tinder but it's only pictures of my wife and I can only swipe right
Show HN: Tinder but it's only pictures of my wife and I can only swipe right
Show HN: Any-LLM – Lightweight router to access any LLM Provider
We built any-llm because we needed a lightweight router for LLM providers with minimal overhead. Switching between models is just a string change : update "openai/gpt-4" to "anthropic/claude-3" and you're done.<p>It uses official provider SDKs when available, which helps since providers handle their own compatibility updates. No proxy or gateway service needed either, so getting started is pretty straightforward - just pip install and import.<p>Currently supports 20+ providers including OpenAI, Anthropic, Google, Mistral, and AWS Bedrock. Would love to hear what you think!
Show HN: Compass CNC – Open-source handheld CNC router
Hey HN,<p>I am Cam, and for the past two years I have been working on Compass, an open-source handheld CNC router that brings computer precision to woodworking while keeping the user directly involved in the process.<p>The idea started as my senior design project at UC Berkeley, with the goal of making a more approachable CNC machine—standard CNC machines are expensive, bulky, and remove you from the tactile “maker” experience. Compass solves that by combining a handheld router with real-time robotic assistance. You move the router roughly along a design path, and Compass uses four optical flow sensors (like in computer mice) and a 3-axis motion system to auto-correct for precision cuts.<p>What is different about Compass:
- Open source: All plans, firmware, and CAD files are available on GitHub.
- Affordable: The DIY build costs ~$600 in parts, and I am selling kits for <$800.
- No external markers: The sensing technology allows for positioning without external markers, so no setup or consumables required.
- Portable: Fits in a backpack and is not limited by a fixed work envelope.<p>We recently completed our first beta program and have just launched V1 kits for pre-order. You can find more info and the launch video at the listed URL.<p>GitHub: <a href="https://github.com/camchaney/handheld-cnc">https://github.com/camchaney/handheld-cnc</a>
Show HN: Compass CNC – Open-source handheld CNC router
Hey HN,<p>I am Cam, and for the past two years I have been working on Compass, an open-source handheld CNC router that brings computer precision to woodworking while keeping the user directly involved in the process.<p>The idea started as my senior design project at UC Berkeley, with the goal of making a more approachable CNC machine—standard CNC machines are expensive, bulky, and remove you from the tactile “maker” experience. Compass solves that by combining a handheld router with real-time robotic assistance. You move the router roughly along a design path, and Compass uses four optical flow sensors (like in computer mice) and a 3-axis motion system to auto-correct for precision cuts.<p>What is different about Compass:
- Open source: All plans, firmware, and CAD files are available on GitHub.
- Affordable: The DIY build costs ~$600 in parts, and I am selling kits for <$800.
- No external markers: The sensing technology allows for positioning without external markers, so no setup or consumables required.
- Portable: Fits in a backpack and is not limited by a fixed work envelope.<p>We recently completed our first beta program and have just launched V1 kits for pre-order. You can find more info and the launch video at the listed URL.<p>GitHub: <a href="https://github.com/camchaney/handheld-cnc">https://github.com/camchaney/handheld-cnc</a>
Show HN: X11 desktop widget that shows location of your network peers on a map
Show HN: X11 desktop widget that shows location of your network peers on a map
Show HN: MCP server for Blender that builds 3D scenes via natural language
Hi HN!<p>I built a custom MCP (Model Context Protocol) server that connects Blender to LLMs like ChatGPT, Claude, and any other llm supporting tool calling and mcps, enabling the AI to understand and control 3D scenes using natural language.<p>You can describe an entire environment like:<p>> “Create a small village with 5 huts arranged around a central bonfire, add a river flowing on the left, place a wooden bridge across it, and scatter trees randomly.”<p>And the system parses that, reasons about the scene, and builds it inside Blender — no manual modeling or scripting needed.<p>What it can do:
- Generate multi-object scenes like villages, landscapes, from a single prompt
- Understand spatial relations — e.g., “place the bridge over the river” or “add trees behind the huts”
- Create camera animations and lighting setups: “orbit around the scene at sunset lighting”
- Respond to iterative changes like: “replace all huts with stone houses” or “make the river narrower”
- Maintain object hierarchy and labels for later editing<p>Tech Stack:
- Blender Python scripting
- Node.js server running MCP
- LLM backend (OpenAI / Claude, easily swappable)<p>Demo:
<a href="https://blender-mcp-psi.vercel.app/" rel="nofollow">https://blender-mcp-psi.vercel.app/</a><p>GitHub:
<a href="https://github.com/pranav-deshmukh/blender-mcp-demo/">https://github.com/pranav-deshmukh/blender-mcp-demo/</a><p>Curious to hear thoughts from folks in 3D tooling, AI-assisted design, or dev interface design. Would you find this useful as a Blender plugin? I’m open to expanding it!<p>Please try it and give it a star on github
Show HN: Conductor, a Mac app that lets you run a bunch of Claude Codes at once
<a href="https://threadreaderapp.com/thread/1945870105109246401.html" rel="nofollow">https://threadreaderapp.com/thread/1945870105109246401.html</a><p><a href="https://x.com/charliebholtz/status/1945870105109246401" rel="nofollow">https://x.com/charliebholtz/status/1945870105109246401</a>
Show HN: Conductor, a Mac app that lets you run a bunch of Claude Codes at once
<a href="https://threadreaderapp.com/thread/1945870105109246401.html" rel="nofollow">https://threadreaderapp.com/thread/1945870105109246401.html</a><p><a href="https://x.com/charliebholtz/status/1945870105109246401" rel="nofollow">https://x.com/charliebholtz/status/1945870105109246401</a>
Show HN: Improving search ranking with chess Elo scores
Hello HN,<p>I'm Ghita, co-founder of ZeroEntropy (YC W25). We build high accuracy search infrastructure for RAG and AI Agents.<p>We just released two new state-of-the-art rerankers zerank-1, and zerank-1-small. One of them is fully open-source under Apache 2.0.<p>We trained those models using a novel Elo score inspired pipeline which we describe in detail in the blog attached. In a nutshell, here is an outline of the training steps:
* Collect soft preferences between pairs of documents using an ensemble of LLMs.
* Fit an ELO-style rating system (Bradley-Terry) to turn pairwise comparisons into absolute per-document scores.
* Normalize relevance scores across queries using a bias correction step, modeled using cross-query comparisons and solved with MLE.<p>You can try the models either through our API (<a href="https://docs.zeroentropy.dev/models">https://docs.zeroentropy.dev/models</a>), or via HuggingFace (<a href="https://huggingface.co/zeroentropy/zerank-1-small" rel="nofollow">https://huggingface.co/zeroentropy/zerank-1-small</a>).<p>We would love this community's feedback on the models, and the training approach. A full technical report is also going to be released soon.<p>Thank you!
Show HN: We made our own inference engine for Apple Silicon
We wrote our inference engine on Rust, it is faster than llama cpp in all of the use cases. Your feedback is very welcomed. Written from scratch with idea that you can add support of any kernel and platform.
Show HN: Shoggoth Mini – A soft tentacle robot powered by GPT-4o and RL
Show HN: I built this to talk Danish to my girlfriend – works with any language
I'm in my 4th year living in Denmark as an expat, and I finally decided it’s time to properly learn Danish. I do have a Danish girlfriend, after all. One way I’ve been practicing is by trying to text only in Danish, but I often find myself stuck. I start my message in Danish, then hit a wall because I don’t know a word or how to fit something naturally into the sentence.<p>Especially in those cases, I used to give up and translate the entire message from English, which kind of defeats the purpose and interrupts the learning process.<p>So I started prompting GPT. I’d write my message with wildcards or notes for the parts I didn’t know, and it would return a corrected version. That worked well, but reusing the prompt each time became tedious.<p>So I built a wrapper around it.<p>Now I can type in the target language, mark unclear parts with curly braces {like this}, and get an instant corrected version with explanations. I also added a history feature so I can review what I got wrong, and I plan to build more on that soon (eg. summary of areas or words to review).<p>This app is for language learners who want to practice writing without feeling insecure about mistakes or breaking their flow by switching to a translator.<p>I hope you find it useful!
Show HN: Bedrock – An 8-bit computing system for running programs anywhere
Hey everyone, this is my latest project.<p>Bedrock is a lightweight program runtime: programs assemble down to a few kilobytes of bytecode that can run on any computer, console, or handheld. The runtime is tiny, it can be implemented from scratch in a few hours, and the I/O devices for accessing the keyboard, screen, networking, etc. can be added on as needed.<p>I designed Bedrock to make it easier to maintain programs as a solo developer. It's deeply inspired by Uxn and PICO-8, but it makes significant departures from Uxn to provide more capabilities to programs and to be easier to implement.<p>Let me know if you try it out or have any questions.
Show HN: Refine – A Local Alternative to Grammarly
Show HN: Refine – A Local Alternative to Grammarly
Show HN: Ten years of running every day, visualized
Today marks ten years, 3653 consecutive days, of running at least one mile every day under the USRSA rules [1]. To celebrate, I built an interactive dashboard that turns a decade of GPX files into charts you can explore.<p>Running has truly changed my life: I've made lifelong friends, explored beautiful places, and more importantly invested into my own health and fitness, which I'm starting to see the positive benefits as I get older.<p>The stack is pretty simple: a NextJS app, with a Postgres database to keep all my running data, and all the stats are pre-computed and cached in Redis, so I effectively only hit the database once a day when a new run is ingested. On the fronted, I toyed with the idea of using D3 or pre-existing data viz libraries, but ended up rolling my own using SVGs directly, it gave me more control on the visualizations.<p>I used the Strava bulk export to pre-populate the database, and I'm using their webhook API to do incremental updates. I have to tap into OpenWeatherMap and OpenCageDate to enrich the running data a little bit.<p>Happy to answer anything about the stack, data pipeline, or how I stayed motivated for 10 years!<p>[1] <a href="https://www.runeveryday.com" rel="nofollow">https://www.runeveryday.com</a> Run Streak Association rules: ≥ 1 mile per day