AI 资讯

Gemini 3.6 Flash ( gemini-3.6-flash ) and Gemini 3.5 Flash-Lite ( gemini-3.5-flash-lite ) are generally available (GA) and ready for production use. Gemini 3.6 Flash : Stronger performance on complex agentic and multimodal tasks while reducing token usage, at a lower price point than 3.5 Flash. Gemini 3.5 Flash-Lite : The fastest, lowest-cost model in the 3.5 family. Outperforms prior Flash-Lite generations for high-throughput execution. This guide explains what's new in each model, what API changes affect your code, and how to migrate. Tip : You can automate this migration with a coding agent that supports skills (like Antigravity). Run /gemini-interactions-api migrate my app to Gemini 3.6 Flash New models Model Model ID Default thinking level Pricing Description Gemini 3.6 Flash gemini-3.6-flash medium $1.50/1M input tokens and $7.50/1M output tokens Balances speed with intelligence for agentic and multimodal tasks. Gemini 3.5 Flash-Lite gemini-3.5-flash-lite minimal $0.30/1M input tokens and $2.50/1M output tokens The fastest, lowest-cost 3.5 model for high-throughput execution. Both models support the 1M token context window, 64k max output tokens, thinking, and the full suite of built-in tools including Computer Use . For complete specs, see the model pages: Gemini 3.6 Flash model page Gemini 3.5 Flash-Lite model page For detailed pricing, see the pricing page . Quickstart from google import genai client = genai . Client () interaction = client . interactions . create ( model = " gemini-3.6-flash " , input = " Write a three.js script that renders an interactive 3D robot. " ) print ( interaction . output_text ) What's new in Gemini 3.6 Flash Token and turn reduction: Completes multi-step workflows with fewer reasoning steps, conversational turns, and tool calls than Gemini 3.5. It also reduces execution loop spiraling. Improved code generation: Produces higher quality production-ready code with fewer unwanted edits and fewer debugging loops. Better instruction f

If you are an enterprise architect working on Microsoft Entra and AI agents, your first overdue job is not another policy wizard, another dashboard, or another governance steering committee. It is schema design. Specifically, it is deciding how you classify non-human identities with custom security attributes in Microsoft Entra . Not eventually. Up front. I keep seeing the same pattern across customers of every size: teams move quickly on agent experimentation, they onboard identities, they test controls, and then they realize they have no consistent attribute language for policy scope. At that point, every policy becomes a naming convention problem in disguise. That is backwards. The control plane starts with classification Custom security attributes are not decorative metadata. They are tenant-scoped key-value classifications you can assign to users, enterprise applications (service principals), and agent identities that are modeled as a service principal subtype, with dedicated role and permission boundaries for who can define and assign them ( overview , Graph model , agent identity service principal model ). That alone should change how architects think about them. This is not "nice to have taxonomy." This is policy input. Microsoft Entra Conditional Access for agents supports attribute-driven targeting with custom security attributes, and policy evaluation happens during token issuance and refresh, not just at policy authoring time ( Conditional Access for agents ). In other words: if your classification is sloppy, your runtime decisions are sloppy. Why agents raise the stakes You can say "an agent identity is still a service principal" and be technically correct. Microsoft Entra Agent ID is built on service principal infrastructure ( agent identities, service principals, and applications ). You can also miss the point. Agent identity introduces a blueprint-centered model where one blueprint can represent many agents, where blueprint-level policy decisions can

Every morning, an online store receives the previous day’s orders from a marketplace partner. The file comes in JSON format. The company needs to add those orders to its main MongoDB orders collection. The sales manager also needs a CSV report that can be opened in Excel. That sounds like a small task. Import the file, copy the documents, export the report. But in practice, a few things can break the process. A date can be imported as a string. A field can have the wrong name. One batch may use total , while the main collection uses totalAmount . A temporary collection can keep old records and trigger duplicate key errors. A CSV export can create null values because the mapping points to fields that do not exist. And then there is customer data. The manager may need the sales numbers, but they probably do not need real customer names or internal customer IDs. This article walks through a real daily workflow: Import marketplace JSON ↓ Store the batch in a temporary MongoDB collection ↓ Copy the orders into the main orders collection ↓ Mask customer fields during export ↓ Create a CSV report The goal is not just to move data from JSON to CSV. The goal is to make the process repeatable, easier to check, and safer to share. The workflow The workflow has three jobs: Import Yesterday Orders ↓ Add Orders to Main ↓ Export Daily Sales Report The important part is the parent relationship between the jobs. Add Orders to Main depends on Import Yesterday Orders , so it only runs after the JSON file is imported successfully. Export Daily Sales Report depends on Add Orders to Main , so the CSV is created only after the main orders collection has been updated. This prevents the report from being generated when data is missing or incomplete. The incoming JSON file The partner sends a file with yesterday’s completed orders. A single order looks like this: { "orderId" : "ORD-2026-07-201" , "customerId" : "CUST-1003" , "customerName" : "Sofia Rossi" , "orderDate" : "2026-07-21T08:20:00

This week's releases cluster around a theme: reducing the overhead that compounds in production agentic systems. Gemini 3.6 Flash ships with measurable token reduction and a price cut, Vercel's AI Gateway gets service tier routing for latency-cost tradeoffs, and Python cold starts quietly drop by half with zero code changes required. Nothing experimental here—most of this is worth touching immediately if you're already in these ecosystems. Gemini 3.6 Flash cuts output tokens by 17% Google's 3.6 Flash reduces output token usage by 17% versus 3.5 Flash while lowering cost to $1.50/1M input and $7.50/1M output. The improvement is most pronounced on coding and web tasks, which happen to be the workload profile of most production agents. The companion model, 3.5 Flash-Lite, trades some quality for throughput—350 output tokens/sec—at $0.30/$2.50 per million tokens. Token efficiency isn't a vanity metric in agentic systems. Multi-step workflows compound output costs: every intermediate reasoning step, tool call response, and context accumulation multiplies what you pay. A 17% reduction per model call can translate to significantly more than 17% savings across a full agent loop, depending on how many hops your workflow runs. The throughput number on Flash-Lite matters too—if you're running high-volume document classification or search reranking, 350 tokens/sec opens architectures that weren't cost-viable before. The API swap is a single parameter change. No migration friction, no new authentication surface. Ship it now if Gemini is already in your stack and you're paying attention to inference costs. Replace 3.5 Flash with 3.6 Flash for general agentic tasks; move high-throughput, lower-stakes subtasks to Flash-Lite. Gemini 3.6 Flash and 3.5 Flash-Lite on AI Gateway Both new Gemini models are immediately available through Vercel's AI Gateway, callable via the unified AI SDK with the same cost tracking, failover, and routing you'd use for any other provider. Model selection

Most bibliography failures show up the night before arXiv or the journal deadline: placeholder DOIs, year pasted into volume= , inverted page ranges, invented case reporters. The fix is a paper repo that fails closed from day one . One command pip install https://github.com/SybilGambleyyu/citesure/releases/download/v0.5.68/citesure-0.5.68-py3-none-any.whl citesure init my-paper cd my-paper citesure gate . --preset ci citesure gate . --preset arxiv citesure init writes refs.bib , pre-commit hooks ( gate --preset ci + soft-lint), .github/workflows/citesure.yml , and a short CITESURE.md for coauthors. Empty bibliographies skip hard-ID floors until entries appear. What the gate checks Soft-lint — placeholder number/issue, inverted pages, year-like volume/month/edition, unsafe keys, all-caps titles, missing venues, duplicate DOIs/titles Health — hard-ID coverage floors Promote dry-run — DOIs still buried in url= Live verify — Crossref, doi.org, arXiv, PubMed, Europe PMC, DataCite, OpenAlex, CourtListener Domain packs Fifty-five live-clean packs (demography, sociology, political science, anthropology, ML, law, ecology, …): citesure packs --gate-all citesure packs --run anthropology-classics Evidence: 256/256 integrity · 209/209 claim pairs · 55 packs. Source: github.com/SybilGambleyyu/citesure · Demo: citesure.sybilgambleyyu.workers.dev

Ethereal Bytecode for the Network: Unlocking XDP's Magic! Hey there, fellow tech enthusiasts! Ever felt like the traditional networking stack in your Linux kernel was a bit… sluggish? Like it was taking the scenic route when you needed it to be a supersonic jet? Well, let me introduce you to a superhero that swoops in and turbocharges your network packet processing: eBPF, specifically in the context of XDP (eXpress Data Path). Forget the days of wrestling with complex kernel modules or praying for better hardware offload. eBPF and XDP offer a revolutionary, in-kernel, safe, and incredibly efficient way to program packet processing at the very edge of your network interface. Think of it as giving your network card a tiny, super-smart brain, capable of making lightning-fast decisions before the packet even bothers the main kernel stack. Pretty cool, right? So, buckle up as we dive deep into the wonderful world of XDP and eBPF, demystifying its power and showing you why it's becoming the darling of modern networking. 1. The "What's the Big Deal?" Section: Introduction to XDP & eBPF Imagine a bustling highway (your network). Traditional networking is like having every car stop at a toll booth, get inspected, and then directed by a central traffic controller. This works, but it can get congested. XDP, on the other hand, is like having intelligent on-ramps where some cars can be instantly identified, rerouted, or even rejected before they even hit the main highway. eBPF (extended Berkeley Packet Filter) is the technology that makes this possible. It's a powerful, sandboxed virtual machine that runs within the Linux kernel. Unlike traditional kernel modules, which can potentially crash your entire system if written incorrectly, eBPF programs are rigorously verified by the kernel for safety and correctness before they are allowed to execute. This means you get the power of kernel-level access without the existential dread of a kernel panic. XDP (eXpress Data Path) leverages

An AI can write a perfect email in seconds. Having a real back-and-forth conversation is much harder. Sarah runs a salon. She has an AI assistant that emails her customers when a slot opens up. Last Friday, a customer canceled his booking. The assistant sent an email: "We have an opening tomorrow at 2 PM. Want it?" The customer replied in a minute: "Yes, book it!" The assistant never saw that reply. The slot stayed open. The customer never got a confirmation. This happens more than people realise — not because it's hard to receive email, but because most setups were never wired to close the loop. Sending is easy. Wiring the whole loop isn't. To be fair, receiving and parsing email isn't some unsolved problem — providers like SendGrid, Mailgun, and Postmark have offered inbound email parsing for years. Point your domain at them, and they'll hand you the clean message. But those webhooks only push the message once. There's no inbox to check back later, and no built-in way to link a reply to the right conversation. You have to build that part yourself — and you still can't run any of it on your own servers. There's a second issue too. AI assistants sometimes send a slightly odd reply — nothing harmful, just a little off. Many managed email providers watch for exactly that pattern, and can suspend an account fast. One strange sentence, and Sarah's whole booking system could go dark with no warning. What a real AI assistant needs For an assistant like Sarah's to actually hold a conversation, a few things need to work together: Replies need to land somewhere the AI can read them They need to arrive clean, not messy They need to stay linked to the right conversation The AI needs to reply back from the same email address All of it needs to run on infrastructure you control, not three different vendors This is what we built Reloop for Reloop puts that whole loop in one place, self-hosted. When Sarah's customer replied "Yes, book it!", Reloop caught the reply, cleaned it up,

With “blessings” from the original Jibo founders, iKairos is a wearable or desk-mounted “AI journal” that turns your family moments into AI images and video.

On July 3, as a heat wave swept the region, New York State’s grid imported 52 gigawatt-hours of electricity from Canada—enough to meet about 9% of its total electricity demand that day. Some of that power shuttled in on a 339-mile power line stretching from Quebec to Queens called the Champlain Hudson Power Express (CHPE).…

The bottleneck in a mature SOC is rarely analyst triage; rather, it is the detection-engineering team's ability to keep the rule base aligned with a threat landscape that evolves faster than rules can be written. Learn how multi-agent system for production security operations has reduced mean times to detect and to respond by 40% and compressed the human work required by 12x. By Willem Berroubache

Increase your brand's AI visibility on autopilot Discussion | Link

Permission-first real-time senses for AI companions Discussion | Link

Building Profitable Niche Lifestyle Stores: A Developer's Guide to E-Commerce Essentials Launching a niche lifestyle e-commerce store is more than just setting up a storefront—it's a technical challenge that requires thoughtful architecture, performance optimization, and strategic content planning. Whether you're building a store for outdoor gear, fashion accessories, or humor products like humor24.se , here's what developers need to know. Choose the Right Tech Stack Most profitable niche stores run on WordPress + WooCommerce + a performance-focused theme (like Blocksy). This combination offers: Flexibility : Extend functionality without rewrites SEO-friendly : Built-in structured data support Cost-effective : No complex deployment infrastructure needed Content integration : Blog + products on the same platform Use a headless CMS approach only if you have compelling reasons (high-traffic requirements, complex frontend needs). For 99% of niche stores, the overhead isn't worth it. Core Web Vitals = Revenue Google's Core Updates consistently penalize slow stores. Prioritize: LCP (Largest Contentful Paint) < 2.5s : Optimize images (WebP/AVIF format), lazy load below-fold content, defer non-critical CSS INP (Interaction to Next Paint) < 200ms : Minimize main thread blocking, defer JavaScript CLS (Cumulative Layout Shift) < 0.1 : Use fixed image dimensions, avoid late-loading ads/widgets # Generate WebP variants of product images convert image.jpg -quality 80 image.webp # Check your Core Web Vitals # Use PageSpeed Insights API or Lighthouse CI in your deployment pipeline Each 0.1s improvement in LCP can yield 3-5% conversion uplift. Performance is a feature. Structured Data Wins Traffic Rich snippets dramatically improve click-through rates. Implement: Product schema : name , price , availability , image , brand , AggregateRating BreadcrumbList : Navigation structure FAQ schema : If you have FAQs (47% higher CTR for FAQ snippets) { "@context" : "https://schema.org/" , "@t

Do I Need Ventilation for a Laser Engraver? Safety Facts You Must Know If you're new to laser engraving, you're probably excited about getting your first machine and starting your first project. But here's one question that pops up for every beginner: do you need ventilation for a laser engraver? The short answer is yes, you absolutely need proper ventilation for laser engraving . In fact, it's not just a recommendation—it's a safety requirement. As someone who's been laser engraving for years and has helped dozens of beginners get set up, I've seen what happens when people cut corners on ventilation. Let me break down everything you need to know about laser engraver ventilation requirements, why it matters, and what your options are—even if you're on a tight budget. Why Is Ventilation Important for Laser Engraving? When your laser engraver cuts or etches materials, it doesn't just magically remove material. The laser beam heats up the material to such a high temperature that it vaporizes it. This process creates laser fumes and fine particles that float around in the air. What's in Those Laser Fumes? The exact composition depends on what you're cutting, but here are some common things you'll find: Fine particulate matter (microscopic particles that can get deep into your lungs) Volatile organic compounds (VOCs) that have strong odors and can cause health issues Toxic chemicals depending on the material (formaldehyde from plywood, cyanide from some plastics, etc.) Irritating gases that can make your eyes water and your throat burn Health Risks of Poor Ventilation So what happens if you don't ventilate? Let's talk about the real risks, not just scare tactics: Short-term effects: Headaches and dizziness from breathing in fumes Irritation of eyes, nose, and throat Allergic reactions or asthma attacks Nausea from strong odors Dirty residue covering everything in your workspace Long-term effects: Chronic respiratory problems from repeated exposure to fine particles Incre

TL;DR : most production AI tasks are not LLM tasks. To triage my email, I replaced a 7-billion-parameter model with a tiny classifier in Go. The rule fits in one sentence. Rules first, a small model next, the LLM only as a last resort. The result: no GPU, sub-millisecond inference, and a cloud call that became rare. Here is how, with the real numbers. This article is for developers who put an LLM in production and pay the bill. Not a demo. Most AI tasks are not LLM tasks In 2026, the default reflex is to wire a big model into everything. A question comes in, you call the LLM. But many tasks do not need it. Filing an email under "work" or "newsletter" is classification. A problem solved for twenty years, long before LLMs. To classify is to pick a label from a short, stable list. To generate text is something else. The first job needs a small model. The second earns a big one. The rule I defend fits in one sentence. Put the LLM last. The setup I built an agent that triages my inbox. It is a daemon. It reads new messages and files each one into a category: work, notification, newsletter, promo, and a few more. Nothing secret, just my real mailbox, with years of mail. The first version handed every email to a local LLM. A 7-billion-parameter model, Qwen 2.5 7B, served by Ollama on a GPU. Ollama is a tool that runs an LLM on your own machine. It worked. But the price was heavy. A GPU on all the time. One more container to watch. And an absurd slowness for the question asked. One day I asked myself: does deciding "is this a newsletter?" really need 7 billion parameters? No. The answer sits in two or three words from the sender and the subject. So I rethought the whole thing. Three layers, from cheapest to most expensive Every email goes through three layers, in order. It stops at the first one that can answer. Deterministic rules. Instant, exact, no cost. A small model. Sub-millisecond, on CPU. The LLM. Only if the small model is unsure. The routing code fits in a few lin

As a backend developer, I have build hundreds of endpoints, so the typical endpoint flow is deeply ingrained in how I think about web applications. But when I started building AI-powered endpoints, I noticed an interesting shift. At first, AI endpoints looked like simple proxy endpoints with some configuration for connecting to a model: API receives request ↓ send prompt to model ↓ receive response ↓ return it to the client And it worked well until I found out that passing a prompt directly from the client was not a good idea. The endpoint could be misused for a completely different purpose, allowing someone else to consume my AI usage credits. Then I realized that the input also needed limits. Sending a large context for a specific task costs more and may produce unexpected results. So when I started looking closer, especially when I needed reliable structured output and predictable application behavior, I quickly realized that it was not that simple. Validation was no longer only guarding execution, it had also become a post-processing step. AI models are probabilistic. Even with the same input, they may return different outputs, omit required information, misunderstand instructions or return something that is technically valid but logically wrong. And because every token has a price, I cannot simply retry the request and hope for a better result. That was when I started questioning whether AI endpoints should be designed in the same way as conventional Web API endpoints. Table of Contents Conventional Web API Endpoint Flow AI-powered Web API Endpoint Flow What This Difference Changes Unpredictable Latency Retry Logic Idempotency and Side Effects Testing AI Endpoints The Output Contract Observability and Cost Summary Conventional Web API Endpoint Flow A conventional Web API endpoint usually follows a similar flow: validate request ↓ execute business logic ↓ return representation The first phase is request validation. We validate the incoming data against property

Deploy Python Bots on Render Without Sleep Using StayPresent Render is one of the most popular free hosting platforms for small Python projects, but it comes with a well-known catch: free-tier web services spin down after a period of inactivity and require a fresh incoming request to wake back up. For a render python bot — a Discord bot, Telegram bot, or scraper — that "wake up" delay can mean minutes of downtime every time the service goes idle. This guide covers exactly how Render's sleep behavior works, and how to eliminate it using StayPresent . Table of Contents Understanding Render's Free Tier Why HTTP Port Requirements Matter Setting Up StayPresent for Render Health Checks Explained Self-Ping to Avoid Idle Sleep Crash Recovery on Render Full Working Example Best Practices Common Mistakes FAQs Conclusion Understanding Render's Free Tier Render's free web services sleep after roughly 15 minutes without incoming HTTP traffic. Once asleep, the next request has to spin the container back up before it responds, so real users (or a bot's own polling loop) experience a delay. Render also expects your web service to bind to a port it provides via the PORT environment variable — if nothing is listening there, Render's own health checks will consider the deploy unhealthy. [Render Health Check] --HTTP GET--> [Your Service on $PORT] | No response = unhealthy Why HTTP Port Requirements Matter A typical Telegram or Discord bot doesn't open an HTTP port — it just connects outward to Telegram's or Discord's API and waits for events. That's perfectly normal bot behavior, but it fails Render's expectations for a web service. The fix isn't to change how your bot works; it's to run a small HTTP server next to it, purely so Render has something to check. Setting Up StayPresent for Render Install the production extra so Waitress serves the app instead of Flask's development server: pip install staypresent[prod] Then in your entry point (commonly main.py ): import os import staypres

Track Codex and Claude Code and respond from your Notch Discussion | Link

Every day, I see discussions about how AI assistants and Copilot are changing software development. And honestly? I agree. AI is helping us save time, automate repetitive work, and learn faster than ever before. But recently, I've noticed something that concerns me. Some interviewers, managers, and even developers are starting to treat AI-generated answers as the "correct" answers. That's where I think we're making a mistake. 🤔 Does Copilot Know Your Responsibilities? We've all seen responses like: "With 10 years of experience, you should know this." But who decides that? Does Copilot know: The projects you've worked on? The systems you've built? The challenges you've solved? The responsibilities you've carried for the last 10 years? The answer is simple: No. Two developers can have 10 years of experience and possess completely different skill sets. One may be an expert in distributed systems. Another may specialize in frontend architecture. A third may have spent years building enterprise applications. Meanwhile, a developer with only 5 years of experience may know a modern technology that none of them have ever needed. Does that make anyone less capable? Absolutely not. It simply means their journeys were different. 🚨 Experience Is Not a Checklist Let's take a different example. Suppose someone has spent 10 years mastering Figma and has become an exceptional designer. Does that automatically mean they should be an expert in Photoshop, Illustrator, CorelDRAW, Sketch, and every other design tool? Of course not. Their expertise reflects the work they've done and the problems they've solved. The same applies to software engineering. Experience is about depth, not knowing everything. ⚠️ The Risk of Over-Relying on AI Don't get me wrong. I use AI. Most developers I know use AI. And it saves hours of effort. But there's a difference between: ✅ Taking help from AI and ❌ Letting AI think for you When every answer, every opinion, and every decision comes from AI, something