标签：#ai

This is a cross-post — the original (and any updates) live at broke2builtai.com . The first time I watched Claude Code reach for a skill I hadn't told it to use — read a folder, run the script inside it, and hand back the finished thing — the difference from a slash command finally landed. A slash command waits for you to type it. A skill waits for the situation . Claude decides. That one shift is the whole feature, and building one takes about five minutes once you know where the file goes. Here's the entire thing end to end, including the one gotcha that decides whether your skill ever actually fires. What a Skill actually is A Skill is a folder with a SKILL.md file inside it. The Markdown holds instructions; the YAML frontmatter at the top holds a name and a description . That description is doing the most important job in the whole file: Claude reads it to decide, on its own, whether the current task warrants invoking the skill. Nothing else you write matters if the description doesn't get you picked. That's the mental model to hold onto: a custom slash command is a prompt you trigger by typing /name ; a skill is a procedure Claude triggers when the context matches. Same reusable-instructions idea, opposite trigger. Where the file goes Two locations register, exactly like commands and subagents : Project skill — .claude/skills/<skill-name>/SKILL.md inside the repo. Committed, so your whole team gets it. Personal skill — ~/.claude/skills/<skill-name>/SKILL.md in your home directory. Follows you across every project on your machine. Each skill is its own folder, and the folder name should match the name in the frontmatter. A loose SKILL.md sitting somewhere else won't be picked up. The minimum viable skill Create the folder and the file: .claude/skills/pytest-runner/SKILL.md Then write the two-part file — frontmatter, then body: --- name : pytest-runner description : " Run, generate, or debug pytest tests for this project. Use when the user asks to run the test su

Introduction Chrome's Built-In AI APIs allow applications to perform selected AI workloads directly within the browser. Unlike traditional AI integrations, developers do not need to deploy or operate model infrastructure. This guide walks through the major APIs currently available. Getting Started: API Availability and Chrome Flags Chrome's Built-In AI APIs are at different stages of maturity. Some APIs are available in stable Chrome, while others remain experimental. The required setup therefore depends on the API you want to test. Available in Chrome Stable The following APIs are available in stable Chrome on supported desktop devices: Language Detector API Translator API Summarizer API These APIs do not require experimental flags for normal use in supported Chrome versions. The Prompt API has different availability requirements depending on whether it is used from a web page or a Chrome Extension. Check the current Chrome documentation for the environment you are targeting. Experimental APIs The Writer, Rewriter, and Proofreader APIs remain experimental and may require developer trials, origin trials, or Chrome flags for local development. Because these APIs are evolving, refer to the official Chrome documentation for the current setup requirements rather than relying on a static list of flags. Engineering recommendation: Use feature detection and availability() checks at runtime rather than relying on Chrome version numbers or assuming that a particular flag is enabled. Language Detector API Use cases: Dynamic localization Query routing Analytics Content classification Example const detector = await LanguageDetector . create (); const result = await detector . detect ( " Bonjour tout le monde " ); console . log ( result ); Architecture Notes Low latency Task-specific model Suitable for client-side execution Complete runnable example: Language Detector API on GitHub Gist Translator API Use cases: Localization Offline translation International applications Example

The most interesting model comparison of 2026 isn't a benchmark table. It's a product exec quietly changing the question everyone asks about models — and getting a completely different ranking as a result. Claire Vo (founder of ChatPRD, host of the How I AI podcast) ran a head-to-head between OpenAI's new GPT-5.6 lineup (Soul / Terra / Luna) and Anthropic's Claude Fable and Sonnet. The result was an upset: the most theoretically intelligent model, Claude Fable, lost to the one she could actually collaborate with, GPT-5.6 Soul. Here's what that upset actually reveals. She killed "vibes" — then bet 70% back on her own taste Tired of vibe-checking, Vo built a real benchmark across the work she does every day: writing PRDs, prototyping apps, debugging multi-step code, and talking to an agent. Scoring had two layers — an LLM-as-judge (she picked the harshest judge, GPT-5.5) and her own hand-graded "taste test," where she clicked through every artifact and wrote notes. Then the key move: she weighted the final score 70% her taste / 30% the machine. "It's my show. I trust my own taste more." That's the first insight. Benchmarks are getting more rigorous, but the final call is still human taste. The point of blind testing isn't to replace taste — it's to force it to be honest . Cover the labels, react to the work itself, then put your judgment back at the center. Theoretically brilliant vs. practically effective On raw intelligence, Fable is elite. But Vo's verdict is the sharpest line on models I've seen this year: Fable is theoretically hyper-intelligent. Soul is practically effective. She describes Fable as "an engineer who has never met a human." Precise to the point of pedantry — it scores every risk, hardens every edge. In one case it hardened a tool-calling loop so tightly that only one specific model could run it at all. It optimized itself into a corner. Soul's edge was the opposite: it gets out of its own head. Same stuck problem — she moved it to Codex, said "sto

Over the past week, two narratives have been colliding everywhere I look. On one side, there's panic. AI is expected to replace marketers, engineers, and entire categories of knowledge work almost overnight. On the other, there are quieter but far more consequential signals: enterprise teams discovering their AI infrastructure is burning through API budgets far faster than expected. This isn't because the underlying models are weak, but because the systems built around them are fundamentally inefficient by design. These aren't separate stories. They're the same failure showing up in different places. A conversation with another developer made that gap visible in real time. He argued that auditing a 150,000-line codebase requires feeding the entire repository into a model in one single, massive pass. It's still a common assumption in mainstream tech: that an LLM works like a giant biological brain that you must fully load with raw text before it can begin to think. But that assumption is already outdated. Modern AI systems don't scale through brute-force context. They scale through structure. And that shift changes everything. Key takeaways Bigger context windows did not solve AI. Treating a frontier model as a monolithic processor that re-reads an entire system on every query is wasteful, dilutes attention, and hides bugs under raw volume. ARC-AGI-3 makes the gap stark: frontier models scored under 1% on interactive reasoning tasks that untrained humans solve at nearly 100%. The gap is architecture, not memory. The teams pulling ahead treat the model as one narrow component inside a larger system: intelligent routing, task decomposition, retrieval, and only the minimum necessary context. The next advantage is not the biggest model or the longest prompt. It is the system designed around the model. Prompting was the first generation; systems architecture is the next. The Myth of the Infinite Context Window When context windows expanded into the hundreds of thousands o

OpenAI's long-awaited super app is here.

OpenAI is already shutting down ChatGPT Atlas, its browser that could do tasks for you on your behalf, less than a year after launching it. Atlas was announced in October, but as part of its wave of news about ChatGPT Work today, the company confirmed that it will be "sunsetting" Atlas and is targeting an […]

Preclinical trial is testing the feasibility of humanoid robots in surgery.

You can see if ads on Google Search, Google Discover, and YouTube were made or edited using AI from a new section in Google's "My Ad Center," as reported earlier by TechCrunch. The update, announced on Thursday, adds a "created or edited with AI" label under the "how this ad was made" tab. Users can […]

If Google's AI tools are used, it will automatically be noted in My Ad Center.

Meta's pitch to users is Spark's ability to handle large agentic workloads, fix bugs, and help with large code migrations — the kind of automation that enterprises are increasingly turning to AI companies to provide.

News publishers say OpenAI hid tools and datasets that could identify copyrighted journalism in ChatGPT outputs, escalating their lawsuit with a new motion for sanctions.

canonical_url: https://quantumflow-ai-ecosystem.vercel.app/blog/26-ai-models-compared-2026-cost-guide date: 2026-07-09T10:00:00Z If you're building an AI-powered application in 2026, you have a problem: there are too many models to choose from. OpenAI has GPT-4o. Anthropic has Claude 3.5 Sonnet. Google has Gemini 1.5 Pro. Meta has Llama 3.1. And then there's DeepSeek, Mistral, Cohere, and a dozen others. Most developers solve this by defaulting to GPT-4o for everything. It's the safe choice — powerful, well-documented, and reliable. But it's also expensive: $2.50 per million input tokens, $10.00 per million output tokens. If you're processing 10 million tokens a day, that's $75+ per day, $2,250+ per month. But here's the secret: most of your requests don't need GPT-4o. In this guide, we'll compare 26 AI models across three dimensions — cost, quality, and speed — and show you how intelligent routing can cut your AI bill by up to 90% without changing a single line of your application code. The 2026 AI Model Landscape The AI model market has fragmented into three tiers. Understanding these tiers is the foundation of any cost optimization strategy. Tier 1: Sovereign Local Models (Free, Priority 100-110) These models run on your own hardware (or your users' hardware) via runtimes like Ollama. They cost $0 per token. They're sovereign — no data leaves your infrastructure. They're fast (no network round-trip). And they're getting remarkably good. Model Parameters Context Best For Cost Llama 3.1 70B (Local) 70B 128K Complex reasoning, code $0 Llama 3.1 8B (Local) 8B 128K General chat, fast responses $0 Mistral 7B (Local) 7B 32K Efficient European-language tasks $0 DeepSeek Coder (Local) 6.7B 16K Code generation & completion $0 GLM-4 9B Chat (Local) 9B 128K Bilingual (EN/ZH) chat $0 Llama 3.2 3B (Local) 3B 128K Edge devices, mobile $0 Llama 3.2 1B (Local) 1B 128K Ultra-lightweight tasks $0 CodeLlama 7B (Local) 7B 16K Legacy code tasks $0 GLM-4V 9B Vision (Local) 9B 128K Loca

A new $20 billion valuation would be a giant step up from the $10 billion valuation it reached in October.

A new feature will indicate when advertisers have used generative AI tools to create or edit their ads, Google says.

The Paris-based ElevenLabs competitor, just announced a hefty seed extension round.

There's a pattern I kept seeing. A team gives an agent real capability, like moving money, shipping a change, or resolving a ticket that touches a customer's account. For a while it's great. Then the agent does one thing nobody can explain or defend after the fact, and the entire program snaps back to a human clicking approve on everything. The blocker was almost never the model. It was that there was no clean way to do two things at once. You couldn't bound what the agent was allowed to do before it acted, and you couldn't prove what it did after, in a form that survives contact with an auditor, a regulator, or a customer dispute. You can assemble that from parts today. Use a policy engine to authorize, and an audit log to record. The problem is they're two systems, and two systems drift. Six months later, when someone is actually asking "was this action allowed, and can you prove it," the policy engine and the log disagree about what the policy even was at the time. Now you're reconstructing intent from two sources that were never the same object. That's the gap. Not authorization by itself, and not observability by itself. The thing that authorizes an action and the thing that proves it should be the same object, bound to the exact policy version in force when the decision was made. The primitive Two verbs, one primitive. Control before. You mint a capability, which is a policy scoped to one agent: a spend cap, a counterparty allowlist, an expiry, whatever the action needs. Every consequential action the agent takes gets checked against the committed policy state and returns an allow or deny in the request path. An over-budget or out-of-policy action is refused before it happens, not flagged after. Refused is the operative word. The enforcement point commits no state change for a denied action, no matter how the agent reasons, how it's prompted, or whether it's been compromised. You've turned unbounded irreversible harm into bounded irreversible harm. Prove after

Google AI recently became the official AI Model and Platform Partner of DEV Community. As someone building an AI routing platform, I paid attention. Google's Gemini Enterprise Agent Platform (formerly Vertex AI) promises enterprise-grade AI agent orchestration — and with the DEV partnership, there's never been a better time to explore it. In this article, I'll share how I integrated Google Cloud's Agent Platform with my existing AI router (built on Neon PostgreSQL), what I learned about Gemini's enterprise capabilities, and why the Google AI + Neon + Algolia trifecta is the ideal stack for AI-first applications in 2026. Why Google Cloud's Agent Platform? The Gemini Enterprise Agent Platform is Google's answer to the question: "How do I orchestrate multiple AI agents in production?" It provides: Pre-built agent templates for common workflows (customer support, code review, data analysis) Grounding with Google Search — your agents can cite real, current sources Context caching — reduce costs by reusing conversation context across turns Multimodal understanding — Gemini processes text, images, audio, and video in one call Enterprise security — VPC controls, data residency, IAM integration For QuantumFlow AI (my AI routing platform), the Agent Platform solved a critical problem: how to orchestrate 26 different AI models without building a custom orchestration layer from scratch. The Architecture: Google Cloud + Neon + Next.js Here's the stack I built: User Request → Google Cloud Agent Platform (Gemini orchestration) → QuantumFlow Router (selects optimal model) → Local models (Ollama — free, sovereign) → Cloud models (GPT-4o, Claude, DeepSeek, Gemini) → Neon PostgreSQL (logs, analytics, cost tracking) → Algolia (search across all AI responses) Why Neon (DEV's Database Partner)? Neon is dev.to's official database partner, and for good reason. It's serverless PostgreSQL with: Database branching — create a full database copy in seconds (like git for data) Bottomless storage

Here's the problem nobody talks about: the reason most AI outputs are mediocre isn't the model — it's that you asked for a final answer and got one. A model with no friction produces the path of least resistance. It pattern-matches to "good-enough" and stops. It doesn't know what your bar for quality is. It doesn't know what logic you'd push back on, what tone would make your audience tune out, or what structural flaw a sharp reader would catch in the first 30 seconds. It just fills the token space with the most statistically probable response and calls it a day. So the output hits your clipboard. You read it. You sigh. Then you spend 40 minutes editing something that should have come out right the first time. There's a better way — and it exploits the fact that AI critique is significantly sharper than AI generation. The Core Insight: Models Are Better Critics Than They Are Authors This sounds counterintuitive, so stay with me. When you ask an LLM to generate something from scratch, it operates in "produce plausible content" mode. The pressure is to fill the blank. But when you ask a model to critique an existing piece — especially if you hand it a specific evaluative persona — it switches into "find the gap between what is and what should be" mode. That's a fundamentally different cognitive task, and it's one where models consistently perform better. Research on iterative self-refinement in LLMs (Madaan et al., 2023) shows that when models are given their own output and asked to improve it with explicit feedback criteria, quality scores improve substantially across writing, code, and reasoning tasks. The key variable wasn't model size or prompt verbosity — it was the presence of a structured feedback loop. The mechanism is simple: the critique generates tokens that constrain and guide the rewrite. Those critique tokens become working context. The model rewrites against them. The output is necessarily better-fitted to the evaluation criteria than anything a single-

"Exactly what that dialog looked like between the government and Anthropic and OpenAI is unclear."

"Show a smile, hide the blade." — The 36 Stratagems, Conceal a Dagger in a Smile Previously on...