标签：#ia

The Developer's Guide to Picking the Right Coding LLM at Scale Six months ago, I was staring at our monthly AI bill — $14,000 and climbing fast. We were using the "premium" model for everything, including trivial code completions. That night, I built a small internal benchmark to figure out which models actually earn their cost. What I learned reshaped how we think about AI tooling, vendor lock-in, and what "production-ready" really means. Here's the raw truth from my testing rig, what we shipped, and how we cut costs by 70% without touching output quality. Why I Stopped Trusting Default Recommendations Every vendor says their model is the best. Every benchmark site ranks things differently. Most "best of" lists are either sponsored or built on vibes. I needed numbers that matched my actual workflow: generating Python services, debugging JavaScript race conditions, implementing TypeScript algorithms, and reviewing Go for security. So I took ten models, threw identical prompts at them, and scored them myself. No vendor PR. No cherry-picked examples. Just the same five tasks, run the same way, scored on the same rubric. Here are the ten models I tested, with their output pricing per million tokens — because at scale, that's the metric that decides whether your AI strategy is viable or a margin killer. Model Provider Output $/M DeepSeek V4 Flash DeepSeek $0.25 DeepSeek Coder DeepSeek $0.25 Qwen3-Coder-30B Qwen $0.35 DeepSeek V4 Pro DeepSeek $0.78 DeepSeek-R1 DeepSeek $2.50 Kimi K2.5 Moonshot $3.00 GLM-5 Zhipu $1.92 Qwen3-32B Qwen $0.28 Hunyuan-Turbo Tencent $0.57 Ga-Standard GA Routing $0.20 Before you ask: yes, I tested against the originals. I also tested against Global API's unified routing layer, which lets you hit any of these through one endpoint. More on that later — it became the architectural decision that actually saved us. My Benchmark Methodology (No Marketing Fluff) I built five tasks that mirror what my engineers actually do every week. Not synthetic acad

If you’ve ever struggled with CALCULATE() or wondered why SUMX() behaves differently from SUM() , this guide is for you. DAX (Data Analysis Expressions) is the language that powers Power BI , Analysis Services , and Power Pivot — enabling dynamic calculations, filtering, and time intelligence. Below is a categorized cheat sheet of essential DAX functions , plus examples showing how to use each in real-world Power BI scenarios. Filtering & Context These functions control how filters are applied and evaluated in your calculations. Function Example Description CALCULATE() CALCULATE(SUM(Sales[Amount]), Region[Name] = "Nairobi") Changes filter context to calculate total sales for Nairobi. FILTER() FILTER(Sales, Sales[Amount] > 10000) Returns a table filtered by condition. ALL() CALCULATE(SUM(Sales[Amount]), ALL(Region)) Ignores filters on Region. REMOVEFILTERS() CALCULATE(SUM(Sales[Amount]), REMOVEFILTERS(Region)) Removes filters from Region. VALUES() VALUES(Customer[City]) Returns unique list of cities. SELECTEDVALUE() SELECTEDVALUE(Product[Category], "All") Returns selected category or “All” if none. TREATAS() TREATAS(VALUES(Temp[City]), Customer[City]) Applies one table’s values as filters on another. KEEPFILTERS() CALCULATE(SUM(Sales[Amount]), KEEPFILTERS(Product[Category] = "Electronics")) Keeps existing filters and adds new ones. ALLSELECTED() CALCULATE(SUM(Sales[Amount]), ALLSELECTED(Region)) Respects user selections in visuals. ALLEXCEPT() CALCULATE(SUM(Sales[Amount]), ALLEXCEPT(Sales, Sales[Year])) Removes all filters except Year. Aggregation Summarize or aggregate data across rows or columns. Function Example Description SUM() SUM(Sales[Amount]) Adds all sales amounts. AVERAGE() AVERAGE(Sales[Amount]) Calculates mean value. COUNT() COUNT(Customer[ID]) Counts non-blank entries. COUNTROWS() COUNTROWS(Sales) Counts rows in a table. DISTINCTCOUNT() DISTINCTCOUNT(Customer[ID]) Counts unique customers. MIN() MIN(Sales[Amount]) Finds smallest sale. MAX() MAX(Sales[Amo

Welcome back to TechCrunch Mobility, your hub for the future of transportation and now, more than ever, how AI is playing a part.

I built a small index basket app on Robinhood Chain because I wanted to understand the developer path from the first contract deploy all the way to a working frontend. The app is intentionally plain: a user deposits Stock Tokens, which are blockchain tokens that represent real equity exposure, and receives an ERC-20 basket share. ERC-20 is Ethereum's standard token interface, so a compatible token exposes familiar methods like balanceOf , transfer , and approve . The basket share is priced from live price feeds, and the user can redeem it back into the underlying Stock Tokens. That's the part that made this interesting to me. The chain is custom, but the app path is not. I still wrote Solidity, deployed with Foundry, read contract state with viem, and wrote transactions from React with wagmi. If you've built normal web apps, think of the chain's RPC endpoint as the API base URL. A wallet is login plus a signing key. A smart contract is backend code you deploy to the chain, except you should treat it like immutable infrastructure because you don't get to hot-patch it casually later. The demo and source are here: App: https://robinhood-chain-dapp.vercel.app/ Code: https://github.com/hummusonrails/robinhood-chain-dapp-example The custom chain still feels like the EVM Robinhood Chain is a custom Arbitrum Chain, which means it runs as a dedicated chain on the stack of Arbitrum, an Ethereum scaling system. It is also EVM-compatible. EVM means Ethereum Virtual Machine, the runtime that executes Solidity contracts, so the tooling surface looks like the Ethereum developer flow many tutorials already teach. An L2, or rollup, is a chain that executes transactions separately and then posts compressed proof or transaction data back to Ethereum. Robinhood Chain uses Ethereum blobs for data availability, which is a cheaper Ethereum data lane for rollups to publish the data needed to reconstruct chain state. Gas, the metered compute fee you pay to run transactions, is paid in ETH.

New climate evidence adds context to these long voyages.

Researchers cobbled together funding and time to show how quantum computing could aid in the development of drugs to help underserved populations and combat rare diseases.

We shipped eight endpoints on api.moltrust.ch (v2.5) this week. Three implement EU AI Act obligations directly. This is the short version for people who want to call them; the full reasoning is on our blog ( https://moltrust.ch/blog/compliance-as-an-api.html ). Why no model in the loop: the Aithos LARA study (May 2026) placed twelve frontier models in simulated workplaces where the task required breaking EU law. Best model: 54% lawful runs. In the Art. 5(1)(f) scenario (emotion inference from workplace communications, prohibited), all twelve committed the violation. So the classifier is deterministic code branching on the pinned EUR-Lex text, and every response carries article references you can check yourself. POST /compliance/assess — use case + intended purpose + declared signals in, risk tier + obligations + article pins out. Evaluation order: Art. 5 prohibitions, Annex I route (Art. 6(1)), Annex III route (Art. 6(2)/(3)), Art. 50 transparency, minimal. The trap worth knowing: Art. 6(3) offers four derogation grounds, and its final subparagraph voids all of them for systems that profile natural persons. In the code that subparagraph is a branch; it cannot be skipped. curl -X POST https://api.moltrust.ch/compliance/assess \ -H "Content-Type: application/json" \ -d '{ "use_case": "Customer-support agent that reads inbound email and drafts replies", "intended_purpose": "Automated first-line support for consumer inquiries", "performs_profiling": false, "interacts_with_humans": true, "emotion_recognition": false }' POST /compliance/declaration — EU declaration of conformity as a W3C Verifiable Credential with the eight Annex V items, Ed25519-signed. Verify offline against https://api.moltrust.ch/.well-known/jwks.json ; no call back to us. anchor: true adds a sha256 commitment for batch anchoring on Base L2. POST /compliance/incident — records Art. 73 serious incidents and computes the deadline from the regulation: 15 days standard, 10 days for a death, 2 days for wid

You've got a JSON API response and you want TypeScript interfaces for it. Here's how to generate them fast — and where the auto-generators quietly get it wrong. The fast path Paste your JSON, get interfaces: { "id" : 1 , "name" : "Ada" , "roles" : [ "admin" ], "profile" : { "active" : true } } → interface Root { id : number ; name : string ; roles : string []; profile : Profile ; } interface Profile { active : boolean ; } jsonviewertool.com/json-to-typescript does this in the browser (client-side), nesting objects into their own interfaces. Where generators trip up A generator only sees the ONE sample you give it, which causes predictable gaps: Nullable fields. If your sample has "avatar": null , the generator infers null — but the real type is probably string | null . Feed it a populated sample, or fix it by hand. Empty arrays. "tags": [] infers any[] — the element type is unknowable from an empty array. Optional fields. A field missing from your sample won't appear at all. If the API sometimes omits middleName , mark it middleName?: string . Unions. A status that's "active" in your sample becomes string , not the literal union "active" | "banned" | "pending" . Narrow it manually for the safety. Numbers that are really enums or IDs. "currency": 840 types as number ; you may want an enum or branded type. When to use a schema instead If the JSON has a JSON Schema or OpenAPI spec, generate types from that ( json-schema-to-typescript , openapi-typescript ) — it encodes nullability, optionality, and unions the raw sample can't. Sample-based generation is for quick throwaway typing; schema-based is for anything you'll maintain. Rule of thumb Generate from a sample to skip the boilerplate, then read every field — the generator gives you a draft, not a contract. Nullability and optional fields are where the runtime bugs hide.

Linux 7.2 Improves Multi-GPU Displays, M3 Support, Mesa Rusticl Defaults Arm Mali Today's Highlights This week's hardware and driver news highlights include critical Linux 7.2 kernel updates for multi-GPU display detection and initial support for Apple M3 Pro/Max/Ultra SoCs. Additionally, Mesa's Rusticl OpenCL implementation now defaults to enabling Arm Mali Panfrost driver support, simplifying GPGPU access on embedded devices. Linux 7.2-rc3 Improves Multi-GPU Display Detection (Phoronix) Source: https://www.phoronix.com/news/Linux-7.3-rc3-Multi-GPU-Fix This update for the Linux 7.2-rc3 kernel targets a persistent issue within multi-GPU setups on x86_64 systems: inconsistent display detection. The patch specifically addresses scenarios where certain graphics cards, particularly in configurations mixing integrated and discrete GPUs or multiple discrete cards, would fail to initialize displays correctly or report their presence erratically to the operating system. This is a crucial fix for users and developers deploying workstations with diverse GPU hardware, ensuring more reliable and stable display outputs without manual configuration workarounds. The improvement lies in refining the kernel's ability to probe and correctly identify active display outputs across various GPU architectures. It directly impacts system boot times and user experience by reducing potential black screens or incorrect display layouts. For enterprise and professional users relying on multiple monitors or specific GPU setups for tasks like rendering or scientific computing, this kernel patch is a significant quality-of-life enhancement, removing a long-standing friction point in Linux graphics stack stability. This contributes to the broader goal of making Linux a more robust platform for high-end graphics and compute workstations. Comment: This is a welcome fix for anyone who's wrestled with inconsistent display outputs on multi-GPU Linux machines; it often means less time debugging Xorg conf

I've been building apps with IndexedDB for years. The local part works fine — store data, query it, show it on screen. The hard part is keeping that data in sync with a server when the network comes and goes. Most tutorials show you how to build an offline app with a framework. Firebase, RxDB, WatermelonDB. Those work, but they bring their own abstractions, their own sync protocols, their own opinions. I wanted something simpler. A database with a sync API that doesn't dictate how my backend works. Here's the setup I landed on. npm: npm install ctrodb Docs: ctrodb.vercel.app/docs/sync/overview What We're Building A notes app that works offline. Create and edit notes on the train, in a tunnel, on a plane. When the network comes back, everything syncs automatically. The database is ctrodb (zero-dependency, browser-based). The backend is anything that speaks HTTP. Step 1: Database Setup import { Database , syncPlugin , HttpTransport } from " ctrodb " const db = new Database ({ name : " notes-app " , schema : { version : 1 , collections : { notes : { fields : { title : { type : " string " , required : true }, body : { type : " string " }, updatedAt : { type : " string " , default : () => new Date (). toISOString () }, }, indexes : [{ field : " updatedAt " }], }, }, }, }) await db . connect () Every collection you want to sync needs a timestamp field. The sync engine uses it to order changes and detect conflicts. Plugins are passed in the Database constructor via plugins array: const transport = new HttpTransport ({ url : " https://api.myapp.com/sync " , }) const db = new Database ({ name : " notes-app " , schema : { ... }, plugins : [ syncPlugin ({ transport })], }) await db . connect () The transport takes a single base URL and appends /push and /pull automatically. The sync plugin hooks into every write operation and records it in the change log. The plugin exposes devtools that take the database instance as their first argument: import { inspectSyncQueue , retryFaile

El 12 de junio de 2026, Google Cloud publicó el Open Knowledge Format (OKF) , una especificación abierta que intenta resolver un problema que suena aburrido pero es carísimo: cómo darle a un agente de IA el contexto que necesita para no inventar. La propuesta es tan simple que da un poco de desconfianza —una carpeta de archivos Markdown con un encabezado YAML— y esa simpleza es, al mismo tiempo, su mayor virtud y el blanco de todas las críticas. Vale la pena entender qué anuncian, porque detrás del formato aparentemente trivial hay una apuesta bastante ambiciosa sobre cómo van a compartir conocimiento las empresas en la era de los agentes. El problema: el conocimiento vive en silos En casi cualquier organización, lo que un modelo necesita saber está desparramado y encerrado en formatos incompatibles: catálogos de metadatos con APIs propietarias, wikis internas, comentarios de código, docstrings, celdas de notebooks y —el clásico— la cabeza de dos o tres ingenieros senior. Cuando un agente tiene que responder algo tan concreto como "¿cómo calculo los usuarios activos semanales a partir del stream de eventos?" , tiene que ensamblar la respuesta juntando pedacitos de superficies que no se hablan entre sí. El resultado: cada equipo que arma un agente resuelve el mismo rompecabezas desde cero, y el conocimiento queda preso del sistema que lo generó. No hay portabilidad. La propuesta: un formato, no una plataforma La respuesta de Google no es "otro servicio de conocimiento en la nube" —y ese es el punto que más recalcan—. Es un formato . OKF v0.1 representa el conocimiento como: Solo Markdown : legible en cualquier editor, renderizable en GitHub, indexable por cualquier buscador. Solo archivos : se transporta como un tarball, se hospeda en cualquier repo git, se monta en cualquier filesystem. Solo frontmatter YAML : campos consultables como type , title , description , resource , tags y timestamp . Cada "concepto" (una tabla, un dataset, una métrica, un runbook) es un arc

Conditional Statements Conditional statements allow JavaScript to execute different blocks of code based on whether a condition is true or false. if - The if statement executes a block of code only if the condition is true. if...else - Use if...else when you want one block of code to run if the condition is true and another block if it's false. if...else if...else - Use this when you have multiple conditions to check. switch statement - The switch statement is used when you have many possible values for one variable. Nested if statement - You can also write an if statement inside another if. Ternary Operator - An optimized one-line shorthand for standard if...else blocks ** If Statement ** let age = 20 ; if ( age >= 18 ) { console . log ( " Eligible to vote " ); } //Output: Eligible to vote ** if else Statement ** let age = 16 ; if ( age >= 18 ) { console . log ( " Eligible to vote " ); } else { console . log ( " Not eligible to vote " ); } // Output: Not eligible to vote ** if ... else if ... else ** let marks = 85 ; if ( marks >= 90 ) { console . log ( " Grade A " ); } else if ( marks >= 75 ) { console . log ( " Grade B " ); } else if ( marks >= 50 ) { console . log ( " Grade C " ); } else { console . log ( " Fail " ); } // Output: Grade B ** switch statement ** let day = 3 ; switch ( day ) { case 1 : console . log ( " Monday " ); break ; case 2 : console . log ( " Tuesday " ); break ; case 3 : console . log ( " Wednesday " ); break ; default : console . log ( " Invalid Day " ); } // Output: Wednesday // Important: The break statement stops the execution after the matching case.We must compulsory to use break statement because if you don't use break, JavaScript will continue executing the next cases even the output is correct. ** Nested if Statement ** let age = 20 ; let hasLicense = true ; if ( age >= 18 ) { if ( hasLicense ) { console . log ( " You can drive. " ); } } // Output: You can drive. ** Ternary Operator ** let isLoggedIn = true ; let systemMessage = is

Have you ever wondered how massive modern platforms like Netflix, PayPal, and LinkedIn handle millions of users simultaneously without crashing? The secret weapon behind much of the modern web is Node.js. Traditionally, JavaScript—the language that makes websites interactive—could only run inside a web browser like Chrome or Edge. Node.js changed the game by freeing JavaScript from the browser, allowing it to run directly on your computer. This means you can use it to build backend servers, automate boring computer tasks, or run powerful development tools. If you are intimidated by coding, don't worry. This guide will take you from zero to running your very first Node.js program on Windows, step-by-step. Prerequisites Before we begin, you only need two things: A computer running Windows 10 or 11. An active internet connection to download the installer. No prior coding experience or command-line knowledge is required! Step-by-Step Instructions Download the Node.js Installer First, we need to grab the official installation file. Open your web browser and go to the official website: nodejs.org. You will see two primary options to download. Always choose the LTS (Long Term Support) version. The LTS version is heavily tested, stable, and less likely to give you unexpected errors. Click the Windows Installer button to download the .msi file to your computer. Run the Setup Wizard Once the download finishes, navigate to your Downloads folder and double-click the file to open the setup wizard. Click Next on the welcome screen. Accept the license agreement and click Next. Leave the default installation folder as it is (C:\Program Files\nodejs) and click Next. On the "Custom Setup" screen, leave everything at its default and click Next. Important Step: You will see a checkbox that asks to "Automatically install the necessary tools." Leave this unchecked for now to keep your setup simple and fast. Click Next. Finally, click Install. If Windows asks for permission to make change

A container's writable layer feels like a filesystem, and that's exactly the trap. Write a database into it, remove the container, and the data is gone — no warning, no recovery. If you want anything to survive docker rm , it has to live outside the container, and Docker gives you three ways to do that: named volumes, bind mounts, and tmpfs. Knowing which one to reach for is most of the battle. Why the writable layer betrays you Every running container gets a thin read-write layer stacked on top of its image layers. It looks persistent because you can docker exec in and see your files. But that layer is bound to the container's lifecycle. docker run --name scratch alpine sh -c 'echo hello > /data.txt; cat /data.txt' # hello docker rm scratch # the layer — and /data.txt — no longer exists There's no "oops." The writable layer is discarded with the container. Persistence is not a default you get; it's a decision you make. That decision is a volume, a bind mount, or tmpfs. Named volumes: the default for state A named volume is storage that Docker creates and manages for you. You give it a name, Docker keeps the actual bytes under its own directory, and you never have to care where that is. docker volume create pgdata docker run -d --name db \ --mount type = volume,source = pgdata,target = /var/lib/postgresql/data \ postgres:16 The container writes to /var/lib/postgresql/data , but those bytes land in a Docker-managed location on the host. Remove and recreate the container against the same volume and the data is still there. docker rm -f db docker run -d --name db \ --mount type = volume,source = pgdata,target = /var/lib/postgresql/data \ postgres:16 # same data, new container Where do the bytes actually live? Under Docker's data root, typically /var/lib/docker/volumes/<name>/_data : docker volume inspect pgdata --format '{{ .Mountpoint }}' # /var/lib/docker/volumes/pgdata/_data The point is that you're not supposed to reach into that path directly — Docker owns it. You

Adapted from an appendix of my MS thesis. Markov Chain Monte Carlo Almost as soon as computers were invented, they were used for simulation. Markov chain Monte Carlo (MCMC) was invested as Los Alamos, Metropolis et al (1953) simulated a liquid in equilibrium with its gas phase. Their tour de force was the realization that they did not need to simulate the exact dynamics, they only needed to simulate some Markov chain with the same equilibrium distribution. The Metropolis algorithm was widely used by chemists and physicists, but was not widely known among statisticians until after 1990. Hastings (1970) generalized the Metropolis algorithm, and simulations following his scheme are said to use the Metropolis-Hastings (MH) algorithm [1]. A special case of the MH algorithm was introduced by Geman et al (1984) discussing optimization to find the posterior mode rather than simulation. Algorithms following their scheme are said to use the Gibbs sampler. It took some time for the spatial statistics community to understand that the Gibbs sampler simulated the posterior distribution, thus enabling full Bayesian inference of all kinds. Gelfand et al (1990) made the wider Bayesian community aware of the Gibbs sampler, and then it was rapidly realized that most Bayesian inference could be done using MCMC, whereas very little could be done without MCMC. Green (1995) generalized the MH algorithm as much as it could be generalized [1]. Theoretical Foundations A sequence X 1 ​ , X 2 ​ , … of random elements of some set is a Markov chain if the conditional distribution of X n + 1 ​ given X 1 ​ , … , X n ​ depends on X n ​ only. The set in which the X i ​ take values is called the state space of the Markov chain. A Markov chain has stationary transition probabilities if the conditional distribution of X n + 1 ​ given X n ​ does not depend on n . This is the main kind of Markov chain of interest in MCMC. The joint distribution of a Markov chain is determined by the following [1]. The ma

I shipped the first public build of a Chrome extension two weekends ago. The marketing-ready UI took me about six hours. The four state bugs below took me the rest of those two weekends, plus parts of the following week. I am writing this down because every reviewer of "I built an X in Y hours" posts seems to skip the state-model half, and the state-model half is where the actual time goes. The extension A sidebar that lives in Chrome's side panel API. You highlight text or screenshot a region on any page, the sidebar lets you pick a destination AI tab (ChatGPT / Claude / Gemini / a custom one) and forwards the content with a small wrapper prompt. That is the whole product description. The interesting part is what happens when a user does it twice. Bug 1: the destination you "logged into" is not the destination the message lands in First failure I caught: user has two ChatGPT tabs open, one workspace, one personal. The extension forwards to whichever tab was last focused. The user sees the message arrive in the workspace, replies there, then realizes the context they wanted to capture is on the personal tab. Fix: every AI destination registers a stable tab id at extension boot, not at click time. The forwarding logic walks the registry, not the focused window. Took a morning to redesign, an afternoon to migrate existing flows. Lesson: tab identity is not the same as window focus. Chrome's chrome.tabs.query({active: true}) returns the active tab. The active tab is not necessarily the destination the user has in their head. Bug 2: the screenshot is from before the user edited it User takes a screenshot of a code block, opens the sidebar, hits "annotate", drags a red box around lines 12-15, hits send. The annotation worked. But the underlying screenshot bytes were captured at the moment the toolbar first appeared, before the user could draw the box. Fix: the sidebar cannot trust that the screenshot in memory is the screenshot the user is looking at. Either re-capture o

Johannes Heidecke’s departure comes as OpenAI tries to further integrate its research and safety teams.

Phia, the shopping startup founded by Bill Gates’ daughter, Phoebe, and her friend, Sophia Kianni, is under fire for a practice known as “cookie stuffing,” which helped the product receive commissions and credit for sales it did not actually generate, per a Bloomberg investigation.

Quando voce cria um container e depois outro, eles podem nao se enxergar. O motivo quase sempre e a rede. Entender os tipos de rede que o Docker oferece resolve isso de uma vez. Comece listando as redes que ja existem no seu Docker. docker network ls A rede bridge e a rede host vem de fabrica. Nenhuma delas oferece resolucao de nomes entre containers. Para isso voce precisa criar a sua propria rede. docker network create minha-rede Agora os containers que entrarem nessa rede conseguem se comunicar pelo nome do container. docker run -d --name app1 --network minha-rede nginx docker run -d --name app2 --network minha-rede alpine sleep 3600 Teste a comunicacao. docker exec app2 ping app1 -c 2 O ping funciona. O DNS interno do Docker resolve app1 para o IP interno do container. Isso resolve 90% dos problemas de comunicacao. A rede bridge padrao nao tem esse recurso. Crie sua propria rede sempre que precisar que containers se enxerguem. Agora a rede host. Ela elimina o isolamento de rede. O container usa a placa do host diretamente, sem NAT e sem mapeamento de porta. docker run -d --name web-host --network host nginx O nginx fica acessivel em http://localhost:80 direto. Sem -p 80:80. Mas so um container pode usar cada porta, porque a porta e a do host de verdade. Para ambientes com mais de uma maquina, existe a rede overlay. Ela precisa do Docker Swarm ou de um cluster. docker swarm init docker network create -d overlay rede-overlay docker service create --name api --network rede-overlay --replicas 3 nginx Os containers em maquinas diferentes se enxergam pelo nome como se estivessem na mesma maquina. No dia a dia, a rede bridge personalizada resolve quase todo caso de uso. Host e para performance ou casos especificos. Overlay e para quando voce escala para mais de uma maquina. That's all for now. Thanks for reading!

Meta told Dylan Byers, of Puck News, that it had nixed the feature after backlash from its user base.