Given the current situation with Layer2, where the event rules keep changing and the prize pool is decided arbitrarily, taking the hit and standing at attention is the right thing to do. But what the community wants is not an apology, but that the rules won't be changed in the middle of the night next time. L2 is about trust, not the speed of announcements. What do you think?

On October 6th, the OKX NOW Global Product and Ecosystem Conference will be held in Singapore. This conference focuses on four main areas: AI, payment finance, trading, and on-chain ecosystem. We sincerely invite all friends to share your genuine insights! @okxchinese @okx and use #OKX达人 #OKXNOW Live reservation link: https://t.co/HeBWiHm0nt

MoE splits the FFN into experts, and each token only goes through the top-2, reducing FLOPs. But during inference, all experts must be loaded into GPU memory, so the real bottleneck is communication and scheduling overhead — you save computing power but pay with bandwidth and latency. This question tests not formulas but whether you have actually run the model. If you think I'm talking nonsense, take a screenshot and keep it.

Gold took a heavy hit in September, dropping 5.9% in a single month, which is considered quite severe compared to monthly performances in recent years. Looking at the market, three clear factors are weighing down gold prices: the U.S. 30-year Treasury yield climbing to 5.62%, pushing the risk-free rate to a high level; the Federal Reserve turning hawkish, with rate cut expectations retreating repeatedly; and the strengthening of the U.S. dollar index. For a non-yielding asset like gold, the opportunity cost of holding it is suddenly magnified, so speculative funds naturally withdraw first. For example, every rise in long-term U.S. Treasury yields leads to a repricing of real interest rate expectations, causing leveraged long positions to reduce holdings, gold ETFs to see outflows, and futures net long positions to decline, all of which can trigger a short-term stampede-like correction. But the truly interesting part is this: even as gold prices are being squeezed hard, central banks continue to buy. China's official gold reserves have reached 2,346 tons, which is not a short-term position but a strategic allocation on the national balance sheet. Global central banks have not just been buying for a month or two: net gold purchases were about 1,082 tons in 2022, around 1,037 tons in 2023, and still over 1,000 tons in 2024, maintaining a high level for three consecutive years. Retail investors are running away, while central banks are picking up—this signal is much more meaningful than daily candlestick charts. Retail investors focus more on sentiment, momentum, and stop-loss levels; central banks focus on monetary credit, reserve diversification, and long-term purchasing power. The two operate on completely different time horizons. My own approach hasn't changed: I don't chase shorts, nor do I rush to bottom-fish. After a sharp drop, the two most feared moves are: first, seeing a breakdown and following with shorts, only to short at a temporary low; second, seeing a rebound and thinking the bottom has been reached, rushing in and getting caught in a bull trap. Especially with V-shaped reversals, which are one of the most common forms of bull traps—first a quick pullback to make those who missed out anxious and bottom-fishers excited, then a second dip. Why rush? Wait for it to fall thoroughly, wait for the trend to reestablish, wait for moving averages, momentum, and capital flows to turn strong again—then adding positions is not too late. Sacrificing some initial gains is exchanged for higher certainty and a steadier holding mindset. Looking at the bigger picture, gold has still risen 8.1% so far this year; the overall trend is not bad. The 5.9% pullback in September looks more like a stress test within the annual trend rather than a trend reversal. The issue has never been whether gold has long-term logic; the issue is whether you can hold on. If your position is too heavy, you can't hold; if your leverage is too high, you can't hold; if you watch the market every day, you can't hold either. So now is more suitable for calm observation: with central bank buying providing support, interest rates and the Fed exerting pressure, and the tug-of-war between bulls and bears, the bottom is often not a single point but a range. What do you think?

$SOL at this position is very likely to break out tomorrow. The resistance above is tight, and once the support below breaks, it will accelerate directly. After so many days of sideways movement, it’s time to give a direction. Every time before the Federal Reserve's decision, the market moves ahead of expectations, and this time is no exception. The fund flow of the Ethereum spot ETF is the most direct signal. Save this judgment first and verify it tomorrow. Keep an eye on #SOL

Romania is building the largest NATO base in Europe, and anti-war voices are being silenced outright. To put it bluntly, once the war is over, arms dealers just move to another place to keep making money, and the locals don't even have the right to say no. Don't ask why; the answer is "European values." Share your thoughts.

Immigration is essentially a test of trust. Entering legally and following the rules is the default contract. I understand there may be hardships behind illegal entry, which is a gray area. But if you kill someone, that proves you are unfit to be a citizen. It's that simple. Maybe I'm wrong, and I'll wait to be proven otherwise.

Nasdaq breaks 9000, retirement accounts follow the rally, but that was back in December 2019. Liquidity is poor in the Asian session, and these historical highs often get pulled back, so don’t rush to chase. When macro data shifts, what rises fast falls fast too. Keep a screenshot and check back later. Keep an eye on it.

The U.S. government created a one-stop platform, stuffing over ten thousand forms into a single entry point. Sounds convenient, but think about who holds the backend data and who defines the "official information." Trump said those who have used AI would understand instantly, but what about those who haven't used AI?

Junior gold stocks are still acting out the script from 2011. Back then, the underground ounce was valued at 10% of the spot price. Now that gold prices have risen so much, the cash flow of large mining companies is so clean it could reflect in a mirror, yet the valuation of junior stocks remains unchanged. If you really calculate at 10%, the underground ounce should be worth $400, a difference of 5 to 8 times. This price gap is either market nonsense or an opportunity. If you have a different view, please share your reasons.