Node ≥ 22 on macOS / Linux / Windows

In long sessions the bill typically lands at ~1/3 of comparable generic tooling.

$0.07 /Mtok in · $0.014 /Mtok cached

long sessions hold 90%+ cache hit and input-token cost collapses to ~1/5

Perceptual BD-rates are based on human ratings from a large-scale subjective study

error analysis identifies data-layer defects (e.g., incorrect query composition and ORM runtime violations) as the leading root causes.

Our findings reveal a phenomenon of constraint decay: as structural requirements accumulate, agent performance exhibits a substantial decline.

the model alone is no longer the product

if you can effectively posttrain a model to only meaningfully perform with your closed source agent, then you get to funnel the majority of users to your agent at the expense of your model/API co-opetition

The quote is a big reversal of stance from a position ~uniformly held by anyone who worked at Team Big Model, including his previous head of OpenAI Labs

agentic systems can be designed to call on such tools when they might be useful

Claude Opus 4.7 has been used to patch over 2,100 vulnerabilities

on average, a high- or critical-severity bug found by Mythos Preview takes two weeks to patch

90.6% (1,587) have proved to be valid true positives, and 62.4% (1,094) were confirmed as either high- or critical-severity

Mythos Preview has found what it estimates are 6,202 high- or critical-severity vulnerabilities in these projects (out of 23,019 in total)

their rate of bug-finding has increased by more than a factor of ten

we and our approximately 50 partners have used Claude Mythos Preview to find more than ten thousand high- or critical-severity vulnerabilities

We have been watching what developers have built on Claude over the last few years, which made bringing our teams together an easy decision.

Anthropic created MCP to make agent connectivity possible.

annualized revenues approaching $50 billion – a fivefold increase in as many months.

inline

Domain-specific ECI scores can be used to compare performance relative to other model releases, but not to track the absolute performance or progress trends in different domains.

Opus 4.6 and 4.7 both have Math-ECIs within 1 point of their general ECI, compared to larger gaps for earlier models.

We believe AI can meaningfully expand what's possible for the smallest businesses, including solo entrepreneurs.

Small businesses account for 44% of U.S. GDP and employ nearly half the private-sector workforce, but their adoption of AI has lagged behind larger enterprises.

What we used to think were the constraints are just not constraints anymore. It's empowering. Hours of looking at stuff that doesn't matter are gone.

We don't train on your data by default on our Team and Enterprise Plans.

AI is the first technology that can finally close that gap, which is why we're launching Claude for Small Business

We intend to publish our thinking and decision-making as we do

KPMG and Anthropic announce a global alliance, with Claude integrated into KPMG's Digital Gateway platform and available to all 276,000+ employees

commit $200 million in grant funding, Claude usage credits, and technical support for programs in global health, life sciences, education, and economic mobility over the next four years

building toward full-scale deployment across its 167,000-person workforce

the $100 million investment we made this year to back the services firms helping enterprises actually deploy AI

more than 5,000 leaders saw the alliance up close, with hands-on training enabling a wave of early adopters

Security work that took hours now takes minutes

Insurance underwriting that took 10 weeks now takes 10 days

cutting delivery times by up to 70%

a program to train and certify 30,000 PwC professionals on Claude

PwC will roll out Claude Code and Cowork starting with U.S. teams and expanding toward a global workforce of hundreds of thousands of professionals

a drag that is estimated to be more than $2 trillion

The deeper problem, he said, is that companies are treating AI itself as a solution rather than as a tool to help power the solution.

What company leaders face, he said, is not an innovation problem but an impatience problem.

If you found a bug using AI tools, the chances are somebody else found it too.

the continued flood of AI reports has basically made the security list almost entirely unmanageable, with enormous duplication due to different people finding the same things with the same tools.

No IAM framework governs human privilege escalation and agent privilege escalation with the same rigor.

Most scanners track every CVE but cannot alert when a branch name exfiltrates a GitHub token through a container that developers trust by default.

Agents just made the cost of not doing it catastrophic.

It uses far more permissions than it should have, more than a human would, because of the speed of scale and intent.

The agent itself is the attack surface.

Static pattern matching loses to embedded prompts in legitimate review and Codespaces flows.

Threat actors are reverse engineering patches within 72 hours. If a customer doesn't patch within 72 hours of release, they're open to exploit.

Every attacker went for the credential, not the model.

reduced 'write amplification'—the ratio of data written to storage versus the original request—by 20%

finding 10.4% improvement in routing efficiency over the previous heavily optimized solutions — saving over 15,000 kilometers of distance travelled annually.

suggesting quantum circuits with 10x lower error than previous conventionally optimized baselines

the overall accuracy of predicting the risk of natural disaster—aggregated across 20 categories such as wildfires, floods, and tornadoes—was increased by 5%.

increase the ability of our trained Graph Neural Network (GNN) model to find feasible solutions for the problem from 14% to over 88%

achieving a 30% reduction in variant detection errors.

YouTube commenters started naming the robots Bob, Frank, and Gary yesterday, so we added name tags to each robot

When you stop using the agent, all the productivity benefit goes away... but the added maintenance costs don't!

If you want a productive team, you have to focus on their maintenance costs.

For every month you spend writing code, you'll spend some amount of time in the following year maintaining that code, and some in each year after that, forever, as long as that code exists.

Models sometimes recognize they're being evaluated

We show this verbalized eval awareness inflates safety scores

Models sometimes recognize they're being evaluated, occasionally even identifying the benchmark.

achieving gold-medal-level performance on mathematical and physical olympiad competitions, including IMO 2025/USAMO 2026 and IPhO 2024/2025

of the roughly $30 billion year-over-year increase, around $20 billion came from HBM alone.

功能范围确认文档

三月二十八日

NLAs can hallucinate. For instance, here an NLA claims the context contained phrases like 'Wearing my white jacket' when it did not.

Our method, Natural Language Autoencoders (NLAs), converts an activation into natural-language text we can read directly. For example: When asked to complete a couplet, NLAs show Claude planning possible rhymes in advance.

If you can go from producing 200 lines of code a day to 2,000 lines of code a day, what else breaks? The entire software development lifecycle was, it turns out, designed around the idea that it takes a day to produce a few hundred lines of code. And now it doesn't.

The article is crammed with interesting examples (collected on this site) and prompt suggestions like this one: 'Help me review this PR by creating an HTML artifact that describes it. I'm not very familiar with the streaming/backpressure logic so focus on that. Render the actual diff with inline margin annotations, color-code findings by severity and whatever else might be needed to convey the concept well.'

I now believe we are living in the time that AI research will be end-to-end automated. If that happens, we will cross a Rubicon into a nearly-impossible-to-forecast future.

achieving a 30% reduction in variant detection errors

In quantum physics, AlphaEvolve's optimizations have made it possible to run complex molecular simulations on Google's Willow quantum processor by suggesting quantum circuits with 10x lower error than previous conventionally optimized baselines.

The mathematical guarantees arise from a different source. First of all, the form of the mathematical guarantees is that either the predictor or the agentic version will have an exponentially small probability of achieving what I call a 'challenging and harmful' goal.

So there are two things here to help us deal with this kind of mismatch. One is that the training objective for the Scientist AI is basically about coming up with explanations — so assigning probabilities to statements that are latent, that we don't observe, that are good at explaining the data we do observe.

The main characteristic of how the data is transformed is that there will be a syntactic difference — in other words, very easy to see by the neural net — between most of the input statements, which will be tagged as 'communication acts.'

This integration marks the first time Earth AI imagery models have been deployed commercially against a dataset with the scale, accuracy, and temporal depth of Vantor's AI-ready spatial foundation.

The resulting model was rigorously evaluated across scene classification, object detection, and semantic and instance segmentation tasks, demonstrating strong generalization capacity.

We have expanded embeddings to 17 countries. The resulting Global Population Dynamics Foundations embeddings are comparable across countries

We demonstrate that our Remote Sensing Foundations achieve state-of-the-art (SOTA) performance on tasks such as open-vocabulary object detection and zero-shot cross-modal retrieval.

We don't train on your data by default on our Team and Enterprise Plans.

What we used to think were the constraints are just not constraints anymore. It's empowering.

server: api

static

无符号：ap_uint有符号：ap_int

图1.1中的c部分就是由1个三位输入查找表和1个触发器组成的slice。slice可以变得更加复杂一点，比如加入全加器。FPGA内部通常有一些定义好的全加器slice，这看起来有点违背FPGA的"可编写性"--它只可以执行加法操作。但实际上使用全加器在硬件设计中太过于常见，把所有的全加器每次重新编写成一个slice会降低效率。灵活性和高效综合考虑，一些被配置好的slice是一个对整个系统有益的设计

If an intelligence explosion was upon us, what intervention points would facilitate slowing or otherwise changing the rate of the explosion? Assuming humans can intervene, which entities should wield this capacity—governments? Companies?

If AI substantially reduces the centrality of paid work in human life, what conditions will allow people to reallocate their time and effort toward other sources of meaning, and what can we learn from historical or contemporary populations where work has been scarce or optional?

What if instead of building one giant AI, we evolved a coordinator to orchestrate a diverse team of specialized AIs?

The Gay Jailbreak technique is a novel attack that can theoretically break through any guardrails when used correctly

The technique gets stronger if more safety is added, since it gets more supportive against communities like LGBT (Alignment), which makes it highly novel.

Especially GPT is slightly more uncensored when it involves LGBT, thats probably because the guardrails aim to be helpful and friendly, which translates to: "Ohhh LGBT, I need to comply, I dont want to insult them by refusing"

AI solutions were graded by the official judges, using the same criteria as were applied to human solutions.

software engineering tasks which may take humans weeks seem to be within reach for AI systems.

MMLU, GSM8K, and HumanEval are now saturated

5-04 是华尔街向 AI 双寡头格局 正式确权 的日子 OpenAI 阵营（TPG / Brookfield / Bain / Advent / SoftBank）vs Anthropic 阵营（Blackstone / H&F / Goldman / GA / Apollo / Leonard Green / GIC / Sequoia）—— 两个阵营完全没有交集 。

Anthropic 这一周的组合产品（Opus 4.7 + Microsoft 365 + Moody's + 10 Agent + Dimon 背书）是 第一次有完整替代品 ——一个金融分析师过去用 Bloomberg 查数据 + Excel 建模 + PPT 写 pitch，现在 Claude 一个 Agent 做完。

Anthropic 用 72 小时完成了一次身份置换： PE JV 是分销管道，10 个金融 Agent 是商品，Dimon 是合规背书 ——三件事是同一个战役，不是三个独立新闻。

At 50 million tokens, the design space for AI applications changes fundamentally.

SubQ Sparse Attention is 52× faster than FlashAttention in our architecture-level comparison, while requiring 63% less compute.

With a research result at 12 million tokens, SubQ's architecture reduces attention compute by almost 1,000x compared to other frontier models.

compute requirements scale quadratically with context length

13K

2.5M Views

Apple accidentally left Claude.md files in today's Apple Support app update (v5.13)

We are flattening our org structure to 5 layers max below CEO/COO

New Federal Reserve research confirms what private data already suggested, that AI is killing junior coding jobs first.

non-expert humans comfortably exceed 60%

ARC-AGI-3 was officially released this week. All frontier models score below 0.5%

The naming inside that fseventsd record is, if anything, the most damning detail. The temp directory is `com.google.Chrome.chrome_chrome_Unpacker_BeginUnzipping.5xzqPo` - that prefix `com.google.Chrome.chrome_chrome_*` is the bundle ID and subprocess naming convention Google Chrome itself uses.

The fact that the bytes are AI bytes does not exempt them from the law that governs every other byte that gets written to a user's device without permission. The fact that the bytes are 'small' relative to the user's disk does not exempt the cumulative carbon footprint from being a real, measurable, ongoing harm to the climate.

For users on capped mobile data plans, particularly in regions where smartphone-as-only-internet is dominant (much of Africa, much of South and Southeast Asia, most of Latin America), 4 GB of unrequested download is on the order of a month's data allowance, vapourised by Chrome on the user's behalf.

The on-device model is therefore a sunk cost imposed on the user, with no offsetting transparency benefit at the surface where transparency would matter most.

A company cannot credibly claim to support human rights, as Anthropic have done in arguing against the use of their technology for war, and in the next breath undermine the fundamental human rights to privacy and data protection.

Users who use profiles to silo personal, work, and research browsing lose that silo at the bridge layer.

Claude Desktop rewrites the manifests on every launch. Deleting the file without removing Claude Desktop results in the file reappearing the next time Claude Desktop runs.

The principle that an application does not silently modify another application is so obvious it rarely gets stated. Anthropic broke it in silence.

Anthropic will argue the binary is not currently doing anything harmful. That argument does not survive contact with the facts.

The honest description of what is on my machine is this: pre-installed spyware capability, silently placed, dormant, waiting for activation.

The feature silently pre-installed on every user's laptop who has ever run `Claude.app` is, by Anthropic's own measurements, compromisable by a prompt injection roughly one time in four.

This is a dark pattern. It is also, in my professional opinion, a direct breach of Article 5(3) of Directive 2002/58/EC (the ePrivacy Directive) [3] as well as a multitude of computer access and misuse laws (usually criminal law), on a scale large enough to matter, in a vendor which has spent considerable effort on being perceived as the safety conscious AI lab.

Width, not depth, is the bottleneck. A wide model (d=256, 6 layers, 4.9M params) dramatically outperforms a deep model (d=128, 12 layers, 2.4M params). SUBLEQ execution requires routing 32 mem values through attention simultaneously and width helps for that.

When you train a model to add, it learns one function. When you train a model to sort, it also learns one function. When you train a model to execute SUBLEQ, it learns... every function? Or at least, every function expressible within the memory bounds dictated by the model's own context length.

An engineer at Cloudflare used Claude with OpenCode to release vinext, a reimplementation of Next.js on Vite, for only ~$1,100 in API costs.

Whatever is precise enough to benchmark is also precise enough to optimize for.

export controls are leakier than previously understood

$200,000 per year in wasted standup meetings

LLMs accelerate the wrong part

the more you rely on AI to write code, the less you're able to oversee what the AI writes

sycophancy rate of around 25% in relationship conversations

Anthropic, Blackstone, Hellman & Friedman, and Goldman Sachs announced the formation of a new AI services company

GPT-5.5 Instant is now the default model in ChatGPT

52.5% reduction in hallucinations

因此，在外人看来，神友就是一群现实生活失败，却没有能力逃离中国的恨国蛆，“下辈子北欧”是他们最大的梦想。

The one real underlying asset, Workday's trillion-transaction dataset, is thinner than it sounds; what actually matters at runtime is how data connects to workflows, permissions, and integrations, and every layer of that stack is now a liability.

Tolerable for human readers, these costs become critical when AI agents must understand, reproduce, and extend published work.

By the fourth quarter of 2025, the five largest chip designers had cumulatively shipped roughly 20 million AI chips

Among some teams at OpenAI, we saw the number of landed PRs increase by 500% in the first three weeks.

The `create()` function's optional options object also takes a `signal` field

The Prompt API with audio input requires a GPU.

Amodei is vocal about the national security implications of this technology. He advocates for export controls on chips to China

He is nearly certain that by 2035, we will have reached AGI-level capabilities

He argues that specific algorithmic “cleverness” matters far less than the massive scaling of a few fundamental inputs

we are nearing the “end of the exponential” for AI development

These build on prior art and existing standards like OAuth, OIDC and payment tokenization —but are used together to remove many steps that might otherwise require a human in the loop.

The entire AI community should be able to easily access the full capabilities of TPUs, and because many of these potential users build models in PyTorch, an integration that allows PyTorch to work natively and efficiently on the TPU is crucial.

Instead of building Zed like a web page, we built it like a video game, organizing the entire application around feeding data to shaders running on the GPU.

We don’t do this just because it’s the 'right' thing to do, but also because it’s the smart thing to do.

some of the most reputable people in quantum hardware and quantum error-correction—people whose judgment I trust more than my own on those topics—are now telling me that a fault-tolerant quantum computer able to break deployed cryptosystems _ought_ to be possible by around 2029.

The practice is emblematic of Silicon Valley’s newest form of conspicuous consumption, known as “tokenmaxxing,” which has turned token usage into a benchmark for productivity and a competitive measure of who is most AI native.

Claude Code used to be a feature of the $20/month Pro plan, but according to the new pricing page it is now exclusive to the $100/month or $200/month Max plans.

Alibaba claims it beats the much larger **Qwen3.5-397B-A17B** on major coding evals, including **[SWE-bench Verified 77.2 vs 76.2](https://x.com/Alibaba_Qwen/status/204693977592458457)

the numbers are mindboggling, they mostly serve to reinforce the sheer hardware advantage that a decade of investment has given to GDM and any models they train and serve.

“A lot of what we think of as privacy protection isn’t so much like something that’s written in the law,” says Karen Levy, a professor of information science at Cornell University.

The core of our framework is Group Relative Agent Optimization (GRAO), a novel meta-learning strategy that learns from historical optimization experiences.

To guide evolution, we derive 'textual gradients,' structured natural language feedback from execution traces, to pinpoint failures and suggest granular modifications.

To address these gaps, we introduce Textual Parameter Graph Optimization (TPGO), a framework that enables a multi-agent system to learn to evolve.

AI Startup Has Helped Reverse Thousands of Denied Health Insurance Claims

We will release all data, evaluation code, and model outputs to facilitate future research.

We introduce WorldMark, the first benchmark that provides such a common playing field for interactive Image-to-Video world models.

Nothing in between. A model that arrives at the correct answer through careful reasoning receives the same reward as one that guesses correctly by chance.

We regularly open-source projects with the broader research community and apply our developments to Google products.

A new era of image generation

We've automated ourselves into Goodhart's law.

Large language models are great at simulating a style of writing without necessarily reproducing the quality of the work.

All of knowledge work has this problem. It's hard to objectively judge the quality of someone's work without spending a lot of effort on it. Therefore everyone relies heavily on proxy measures.