maff

Hi Elon, I would love to find a way to get our perpetual futures integrated into X. It would be a killer product.

trade[XYZ] launched its first market on Hyperliquid 381 days ago. Today, we are joining forces with ICE, parent company of the New York Stock Exchange, to bring perpetual futures on Hyperliquid to U.S. regulated markets. I don’t usually write long posts, but today merits reflection. For years, builders across this industry have worked to show that open, onchain markets could support price discovery, manage risk, and withstand stress to the standards of traditional market infrastructure, all while staying open when and where those markets aren’t. Our achievements stand on the shoulders of those builders. And today, the company that owns the New York Stock Exchange has chosen to build upon the work of every developer, trader, market maker, and user who took a chance on us and our markets. We share this moment with all of you. Markets are shaped by their participants. A new generation of investors trades from their phones, across asset classes, at any hour of any day. Until Hyperliquid, the right infrastructure for those markets didn’t exist. Building these markets was hard. It took novel approaches to pricing, funding, and risk, refined with real capital at stake, in markets that don’t close. It also meant earning the trust of the people trading on them. We’re honored that the community recognized that effort and responded in kind. That trust is what allowed us to grow our volumes to a ~$1 trillion annualized run rate in just over a year, to help pioneer new instruments like pre-IPO perps, and to earn the confidence of storied institutions like S&P Dow Jones Indices. That’s why ICE has chosen trade[XYZ] as its exclusive partner to bring 24/7 perps to the U.S. Shifts in technology have pushed markets toward greater transparency, better price discovery, and broader access: from the coffee house to the curb exchange, ticker tape to the telephone, the trading floor to digital exchanges. ICE helped write that last chapter. In 1997, Jeff Sprecher bought a small power-trading technology business for $1. ICE launched as an electronic energy marketplace in 2000, helped digitize major commodity markets, and acquired the NYSE in 2013. It has grown exponentially because of this appetite for innovation. That’s the partner we wanted: a partner that looks forward and embraces progress. Americans deserve access to the most innovative markets in the world, and the protections of U.S. regulation. With trade[XYZ] and ICE, subject to regulatory approvals, they’ll have both. The difference isn’t just access. Perps don’t expire, so traders never have to roll positions, and liquidity concentrates in a single deep market for each asset. Cross-margining enhances capital efficiency by letting one pool of collateral back positions across stocks, commodities, and crypto. Underpinning it all are atomic clearing and settlement, real-time risk management that enables continuous markets, and universally applied rules that anyone can verify. This is the next chapter we’ll write together. My responsibility, our responsibility, is to now build markets worthy of the confidence our partners and our users have placed in us. Our commitments will not waver: to fair and transparent markets, to diverse markets, to low costs of execution, and above all, to market integrity. All of this in pursuit of one thing: A universal exchange, open around the clock, coming to the United States. We’re proud to have deployed more than 100 markets on trade[XYZ]. But measured against the opportunity ahead, we’ve barely started. There is an enormous universe of assets we haven’t reached, alongside things people want to trade that have no established market at all. Wherever there is demand to trade an asset, we want trade[XYZ] to be where its price is discovered. Global stocks, commodities, bonds, pre-IPO companies, compute, new indices, digital assets, interest rates, equity baskets, volatility products, alt assets… Anything. Anytime. Anywhere. Ultimately, everything you own should work with everything you trade. Picture a headline breaking on Saturday night. You should be able to post tokenized shares as collateral, hedge your portfolio with an index perp, add exposure to oil, and take a view on a company that has not yet gone public. All in one account, all on one risk engine, before any traditional market opens. trade[XYZ] and Hyperliquid were built for this. I hope someone opens their first investment account years from now to find that trivial, and to wonder why their predecessors needed so many separate financial relationships. Getting there will take liquidity, reliability, trust, and prices people have reason to believe. It will take years. I’m prepared to spend them. Finally, a few thank yous. First to the XYZ team: For the tireless nights and for committing to this vision before it was obvious. To Jeff Sprecher and ICE, with special thanks to Michael Blaugrund and Trabue Bland who championed us throughout: You’re a world-class, technology-embracing team that proves innovation and scale are in fact not at odds. To Stan Druckenmiller: For your partnership and helping us turn the next page. To Hyperliquid Labs: For being the best partners we could ask for. You set an impossible standard, one we continually strive to meet. @chameleon_jeff, @iliensinc, @xulian_hl And most importantly, to our users: You traded these markets when no one thought it mattered. Guess what? It did.

Very likely Fable 5.5 found in FrontierCode 2.0: Cognitions data reveals a hidden model, "Obsidian 3," scoring 46.8% against #1 Opus 5.5’s 45.5%, and since it runs in the claude-code harness used exclusively by Claude models, its save to say its Fable 5.5.

bio全力で寄り付き買い出した🙏

Hi all, I have a new result I'm excited to share. It turns out that *all existing implementations* of softmax and cross-entropy (that's JAX, PyTorch, textbooks, etc) have a subtle bug --- two bugs, actually. A simple rewrite fixes both, and eliminates "slingshots" when grokking:

Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to browse, search, and use applications. On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments. While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software. After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. We’ve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely. openai.com/index/securing-ou…

【サービス終了のお知らせ】 「僕の魔界を救って！」は、深刻なPHP脆弱性への対応が困難となったため、サービスを終了いたします。 突然のお知らせとなりましたことを、心よりお詫び申し上げます。10年以上にわたり遊んでくださった皆さま、本当にありがとうございました。 #bokumaka

maffさんドカドカと張ってましたよ。公表組なら日本人最大じゃないかな。 ほんま可哀想😢

サイクルが速すぎるゆえに他の時間をかけて形成されるものが間に合わない可能性もある、tokenの価格やコミュニティなど またpapertradeの流動性はHyperliquidの流動性等に制約されるので効率が良くてもスケールしないという問題もある

para:PAPER has been halted and settled. The @papertrade_xyz launch plan, published October 1, included a phase in which PAPER would become transferable. para:PAPER was listed as a pre-launch perp on that basis, with the expectation that it would convert to a standard perp once a spot market became available. Papertrade announced today that they do not intend to make $PAPER transferable. The pre-launch perp relied on the expectation that its underlying will eventually trade. Traders took positions based on a public roadmap that supported that expectation. With it removed, there is no future spot price for the market to converge to. We halted para at 22:31:15 UTC immediately following the announcement rather than schedule a later closure. This established a common cutoff for all open positions and a consistent basis for settlement, limiting the impact of trading around the announcement. All open positions were settled at $0.026296, calculated as the time-weighted average oracle price over the hour ending at the halt, using the same one-hour averaging window as Hyperliquid’s validator-led delistings for HyperCore perps. All open orders were cancelled. Remaining margin has been released, and realized PnL is reflected in account balances. No action is required from traders. Should PAPER’s transferability plans change, we will review whether the market can be unhalted.

Do you believe in forever? enterforever.com