Sonnet 4.6 Max (feb 2026) became the good enough for common tasks baseline. Best win for most companies today is having open, reliable and most importantly cheaper models reaching those benchmarks and to continue exceeding it to keep pressing costs down.
Congrats on the soft launch. Small request, if I do not have a seatbelt on and look away for too long, can it autoplay Stayin' Alive from the Bee Gees on full blast?
We're opening the waitlist for our Monetization Gateway, which will allow you to charge for any web page, dataset, API, or MCP tool behind Cloudflare. The charges will settle in stablecoins over the x402 open protocol. https://t.co/pvICtEIixj
CHINA CONSIDERS RESTRICTING OVERSEAS ACCESS TO CUTTING-EDGE AI MODELS
China’s Ministry of Commerce has led meetings over the past month with major AI companies, including Alibaba, ByteDance, and https://t.co/YDe0KRldDB, to discuss measures that would restrict overseas access to cutting-edge AI models, including models that have not yet been released.
The discussions reportedly include not only closed-source models but also open-weight models. However, the scope of application is still under debate, and the rules may ultimately apply only to future frontier models.
Officials have also discussed designating the leakage or theft of proprietary AI technologies as a national security crime, with stronger penalties, as well as restricting the types of foreign capital that can invest in Chinese AI startups.
The backdrop is the U.S. move to strengthen export controls on AI models, along with national security concerns over cutting-edge models that could possess advanced cyberattack capabilities.
Chinese authorities are reportedly concerned that advanced U.S. cybersecurity AI models could be used to exploit vulnerabilities in Chinese software.
Since the beginning of this year, China has continued to tighten measures to prevent AI technology from being transferred overseas. Authorities have investigated whether Chinese AI startups that relocated abroad violated export control laws, while also strengthening oversight of overseas transactions involving Chinese investors, technology, data, and national security concerns.
Future regulations could take the form of a tiered framework based on technological capability. Basic open-source AI models may be managed through a filing system, high-performance models may be subject to security reviews, and the most sensitive frontier models may be banned from public release or restricted to use within China.
No intelligence for you unless you're the select few? Looks like we are officially entering a new era of gov. crackdowns. First it was treating cryptography for centuries like a military weapon, now its advanced intelligence. History does not repeat, but it rhymes.
As a result of a US government directive, we are suspending access to Claude Fable 5 for all users. You can continue to use all other Claude models.
Here’s what this means for you:
Across Claude products, new sessions will run on your selected default model or Opus 4.8, and existing Fable 5 sessions will end with an error.
On the Claude Platform, requests to Fable 5 will also return an error. Please update your integrations to other Claude models.
We know this is a disruption to your workflows; we appreciate your patience and support.
Sonnet 4.6 Max (feb 2026) became the good enough for common tasks baseline. Best win for most companies today is having open, reliable and most importantly cheaper models reaching those benchmarks and to continue exceeding it to keep pressing costs down.
The US government spends ~$1T every ~100 days ($39T in debt). You'd think universal child care for 3-4 year olds (est. ~$20-35B/yr, fraction of spend!🤔) would be easy by fixing priorities instead of a wealth taxing unrealized gains that distorts markets. Scapegoating 101
Introducing GLM-5.2: Frontier Intelligence, Open Weights
- Significant improvements in coding and agentic tasks
- Strong long-horizon capabilities with a 1M context window
- Two levels of reasoning effort: GLM-5.2 (max) pushes the limits, while GLM-5.2 (high) strikes a strong balance between performance and token efficiency
- MIT-licensed open weights
- Same API pricing as GLM-5.1
Tech Blog: https://t.co/LAsxUdN0JZ
Weights: https://t.co/g0A1C4UWx4
API: https://t.co/Kc3E22cbN7
Coding Plan: https://t.co/Nk8Y98HNhU
Chat: https://t.co/WCqWT0qCQb
Google's IPO created roughly 1,000 millionaires. Facebook's did about the same. SpaceX just created 4,400, with 400 of them above $100 million each.
The structural reason matters more than the feel-good stories.
Companies that stay private for decades usually concentrate the upside in VCs and founders. By the time regular employees get liquid, the 100x already happened on someone else's cap table. SpaceX stayed private for 24 years, the longest run to a mega IPO in history, and routed that compounding into payroll. Stock grants at every level, plus share purchases through paycheck deductions, down to welders and cafeteria staff.
Run the math on what that meant. SpaceX was valued around $12 billion in early 2015. The IPO priced it at $1.77 trillion. That's 147x in 11 years, roughly 57% annualized. A welder buying shares with $200 paycheck deductions compounded at a rate most funds never touch once.
The part people gloss over: this was financing, and the workforce carried real risk. In 2008 the company was weeks from bankruptcy after three failed Falcon 1 launches. Below-market salary plus rocket startup equity was a terrible deal in every timeline where the rockets kept blowing up. For two decades the employees were, collectively, one of SpaceX's largest investors.
Thousands of startups pay in equity. Almost all of that paper expires worthless. This bet only paid because the company survived 24 years, landed the rockets, and went public at the largest valuation in IPO history.
4,400 people just got paid for holding through every one of those filters. The factory floor was the venture capital.
Musk selling the shovels, while having a stake with grok and cursor? Nothing seems to stop that guy his ambitions. Greatest entrepreneur ever and first trillionaire in the making, called it a decade ago while his net worth was ~10bil!
@elonmusk posted this almost a year ago:
"Step 1: Buy a sh*tload of GPUs
Step 2: ?
Step 3: Profit"
Today we found out what Step 2 actually was.
@Google and @AnthropicAI are now paying @SpaceX $2.17 billion per month combined just to rent compute capacity.
He wasn’t joking. He was telling us the plan in meme form.
The man really does think in decades while everyone else is stuck in quarters.
Remember that all government money is counterfeit.
Such money began as discrete quantities of a real commodity: silver or gold. Such money was used and trusted because it was a bearer proxy for this real good; the paper (valueless) a receipt for a quantity of metal (valuable).
And then the metal was stolen by the US Federal Government from all holders of the proxy instrument. The public was rugged, and now all government money is counterfeit, aka fiat.
That the world continued on without rebellion says something about how easily the masses may be deceived so long as a charismatic man with a flag pin stands before them.
When will he next appear before us? What illusion will we next be asked to believe?
The bill for this crime is paid every year by everyone holding fiat, as they see prices rise and struggle with the attendant consequences of being paid in an asset debased in perpetuity. They attribute their suffering to some phenomenon of nature, or to the greed of capitalists, rather than to that specific act of fraud on August 15 1971.
Very predictable, the real worry is capital flight. Guess who wins from deposits staying put and no private competition and no innovation? Banks.. they keep control and keep profits, EU citizens losing open finance over so called 'possible stabilty concerns'
Stablecoins are not an efficient way to strengthen the international role of the euro, says President Christine @Lagarde.
The best solution remains deeper capital market integration through the savings and investment union and a stronger safe asset base https://t.co/Xewr8ysz9B
Meet Kimi K2.6: Advancing Open-Source Coding
🔹Open-source SOTA on HLE w/ tools (54.0), SWE-Bench Pro (58.6), SWE-bench Multilingual (76.7), BrowseComp (83.2), Toolathlon (50.0), Charxiv w/ python(86.7), Math Vision w/ python (93.2)
What's new:
🔹Long-horizon coding - 4,000+ tool calls, over 12 hours of continuous execution, with generalization across languages (Rust, Go, Python) and tasks (frontend, devops, perf optimization).
🔹Motion-rich frontend - Videos in hero sections, WebGL shaders, GSAP + Framer Motion, Three.js 3D.
🔹Agent Swarms, elevated - 300 parallel sub-agents × 4,000 steps per run (up from K2.5's 100 / 1,500). One prompt, 100+ files.
🔹Proactive Agents - K2.6 model powers OpenClaw, Hermes Agent, etc for 24/7 autonomous ops.
🔹Claw Groups (research preview) - bring your own agents, command your friends', bots & humans in the loop.
-
K2.6 is now live on https://t.co/YutVbwktG0 in chat mode and agent mode.
For production-grade coding, pair K2.6 with Kimi Code: https://t.co/uvoSJKyGCY
-
🔗 API: https://t.co/EOZkbOwCN4
🔗 Tech blog: https://t.co/9wWvgIQSS3
🔗 Weights & code: https://t.co/Be0hjs2RTP
Agent skills, no common registry, no lockfile, no version pinning. Just vibes and copy-paste Anthropic plugins, repos, and Vercel npx skills into other repos? "They're non-deterministic"? Maybe not semver per se, however it feels like we need it and we just have not done it yet..