麒麟送财

#alpha 7月2號，Alpha空投預告！人數12.2W！
📅 今日空投
盲盒空投，226分，18:00，盲猜老幣，約30U。
昨天限價單總成交量幹到約16.32億美元，較前一天下降了8.41%。
Binance Alpha 交易賽：
- KGEN：今晚21點就結束了！昨天門檻還是158500，今天直接衝到221578，一天漲了63078名。
- STABLE：這個真離譜，昨天才260，今天飆到47418，暴漲47158名，坐火箭了屬於是。
- ARX：也瘋了，昨天17302 → 今天92434，一天衝上去75132名，手速都太快了。
今日推薦（30天內上線幣種，積分×4）
純交易量推薦ARX，還剩20天。每筆約200-500U，小額多筆刷。

Recently I was cooking at home and discovered an interesting phenomenon: when I follow the recipe strictly, the flavor is consistent, but I always feel something is missing—like it lacks a soul. When I loosen up and freely improvise, it’s likely to turn out badly. This contradiction of “needing both rules and inspiration” becomes even more obvious in investing. Especially when I started looking into the AI trading track, I found that most tools in the market are black boxes—strategies can’t be verified, performance is all based on what people say, great projects get buried, and the ones that cut corners and profit off others’ losses climb to the top through marketing.
So when I saw @NewtonProtocol , my eyes lit up. What it wants to do is essentially provide trust infrastructure for these fast-ramping AI strategies—making automated robot strategies transparent and verifiable on-chain. Specifically, store AI parameters and backtest results on-chain; developers speak for themselves with real performance; users can invest with confidence. Before every trade, the system performs strategy validation—if it doesn’t meet the rules, it’s intercepted immediately and never reaches the settlement layer. After execution, it generates an attestation with a cryptographic signature, so anyone can independently verify it. For price risk control, it integrates RedStone oracles: if the collateral price crosses a threshold, it automatically blocks or liquidates the position, with no manual intervention throughout.
So how does it run? The foundation is EigenLayer AVS, leveraging Ethereum’s economic security. The strategy language uses an enterprise-grade Rego standard—clearly preparing from the start for institutional interfaces. Currently in the Mainnet Beta phase, the VaultKit SDK logic is already able to run through; cross-chain rebalancing and risk-control related fees are paid with $NEWT . The token supply is fixed, and the mechanism also accounts for long-term game theory.
Of course, Beta is still Beta—the real stress test hasn’t arrived yet. Node participation has hardware and registration thresholds, so ordinary users are more like coin-holders who observe for now. The market cap is still low, so whether the mechanism can withstand real traffic is the most critical point to watch.
Honestly, with AI + Crypto being this hot in 2026, I recognize the pain point Newton hit. It’s like a new reagent bottle in a lab: the concept is solid, and the on-the-ground implementation also has real substance. I’ve already added it to my watchlist. Let the “bullets” fly for a bit—let the on-chain data do the talking. If you’re interested, you can try it on the Beta yourself and experience firsthand whether the pre-trade interception plus on-chain credentials actually feels smooth. #Newt

July 1st, #alpha 7 Alpha Airdrop teaser! 12.4W participants!
📅 Today’s airdrop
There’s no airdrop for now today, but! Tomorrow, July 2nd, most likely around 16:00 there will be an old-coin WDATAIP airdrop. If you’re one of the brothers who participated in that IP token swap contract before—set an alarm, don’t miss it.
Yesterday’s limit order trading volume was $1.782 billion, up more than 7% from the day before. The market is clearly recovering—keep up your speed, don’t get left behind.
Binance Alpha Trading Competition:
- STAR: Ends tonight at 21:00! Yesterday had 6,431 participants—today jumped straight to 71,184, an increase of over 60,000 in one day.
- KGEN: Yesterday 102,981 → Today 158,500, up by about 55,000+ people—solid as a rock.
- ARX: Yesterday 0 → Today 17,302, up by 17,000 in a day—new coins are fierce.
Today’s Alpha recommendations (to be launched within 30 days, points ×4):
Pure trading volume recommendation: ARX (21 days remaining). Do each trade around 200–500U, and use many small orders.
Also, a free points mission is here for +5 Alpha points:
It’s enough to predict that the market’s single buy will be greater than 50U. The slippage/loss is about 0.1U, basically like free farming.
The activity ends on July 7th at 7:59 AM. Path:
1. Go to the Binance Alpha event page and click the task entry
2. In categories, choose “Culture”
3. Find this market: “Will the Second Coming of Jesus Christ occur before 2027?”
4. Choose “No” (don’t pick the wrong one)
5. Market buy over 50U, then immediately market sell it—get 5 points (arrives the next day).

Last week I used an AI trading bot to automatically farm airdrops. It ran all day, and in the end I couldn’t even find out what operations it actually performed in the middle. The answer, I guess, is “model decisions,” and if you ask again it’s “a black box.” That feeling of wasting money and burning time, with not even an explanation, is really frustrating.
So the past two days I’ve been seeing Mainnet Beta with @NewtonProtocol , and honestly my first reaction was a bit confused. When other projects launch, they first throw performance numbers at you—how high the TPS is, how fast and powerful. This one, however, right out of the gate puts the validation mechanism in your face.
At the time I thought, “Isn’t that just asking for no one to be impressed?”
But after flipping through the execution flow and validation logic back and forth several times, it slowly clicked. They weren’t trying to outdo everyone on performance. What they’re “competing” for is trust. It’s not “run the task first and then do a check later.” Instead, it validates as it executes, and every step leaves an on-chain record so you can’t wriggle out of it. The deeper you think about it, the more unsettling it is—in a good way: for AI Agents, the black box era is over. Operations are fully transparent, and users can even set an upfront capital limit to directly block permission-overreach risk.
While reading the docs I also found something pretty practical: they didn’t do a one-size-fits-all approach. They really stress-test the critical parts, and they lightweight the non-critical parts. Resource allocation is very clear-headed. This kind of trade-off is far more real than mindlessly piling on computing power.
Right now I’m watching $NEWT too. I’m not really looking at short-term price swings—I want to see whether this layered validation mechanism can truly stand firm in the Mainnet Beta. If they can genuinely solve the trust pain points of AI Agents, and more projects dare to connect smart wallets and automated trading, then that’s real value.
Of course, any new thing has to go through a period of adjustment. Cross-chain and operations/maintenance are currently consuming NEWT, and the total supply is fixed with no additional issuance. The model is fine—no issues there. Compared with those black-box projects with operations you can’t trace, Newton at least has verifiable behavior committed to the chain. I agree with this direction.
I’ll keep tracking the run numbers in Mainnet Beta. If you’re a brother looking for long-term value in AI + crypto, focus more on real network data and less on chasing FOMO. $NEWT ’s road is still long. What’s truly valuable might be this latest attempt at trust at the protocol layer. #Newt

newt $NEWT
A few days ago, a friend complained to me. He said he’d made some quick money using an AI trading bot, but before he even got to enjoy it, he found that his funds were locked in a risky contract—caught between entry and exit, with no good options. He laughed bitterly and said: the robot runs faster than the mind, and the speed at which it loses money is also faster than the mind.
That saying really stuck with me. To be honest, AI trading is indeed booming right now—millisecond order snatching, automatic compounding, and humans simply can’t keep up. But the problem is: AI only follows code instructions. It doesn’t care about address safety or compliance risk. Once you step into a trap, by the time you realize it, your money is already “cold.”
Following that line of thought, I started looking for projects that are addressing this pain point. Later I came across @NewtonProtocol and saw that it’s exactly what it’s working on—making AI run fast while still obeying the rules.
It doesn’t take the traditional manual approval route, because that speed can’t possibly match AI. Instead, it uses a programmable strategy engine, with authorization rules written into the code—for example, only authorized, unexpired delegations are allowed to move funds. In this way, no matter how reckless the AI tries to act, it still has to follow this logic. It’s essentially like putting a compliance layer of guardrails on the robot.
The $NEWT token is crucial in this mechanism, but it’s not used for voting. It’s used as an economic security deposit for operators. Every time a strategy is validated, it consumes a bit of the tokens. If an operator approves a problematic transaction, the staked assets get slashed. Put simply, this design makes mistakes painful—so nobody would dare to treat real money like a joke.
At the time, I tried connecting my wallet to Newton’s vault to catch an RWA arbitrage opportunity. The result was that the trade got stuck in a pending state. The operator network was slowly evaluating the strategy, and by the time it was finally approved, the window had already passed. Honestly, it was a bit disappointing—but on second thought, it’s slower, yes, but at least the money is safe. That feeling of AI charging in unfiltered, only to wake up and find your funds locked—I don’t want to experience that.
Now Agentic Finance is getting more and more popular. AI autonomous trading isn’t just about speed anymore—it also needs a verifiable compliance layer as a backstop. Newton’s approach, which ties the strategy engine, operator network, and on-chain/off-chain data together, is essentially adding a fair authorization gate to DeFi. Of course, there’s always room for iteration with something new—but at least it ensures AI trading isn’t a “bare-bones sprint” anymore. It runs with strategy guardrails. #Newt

After being in the crypto world for a while, looking at projects too long starts to feel like a professional illness. First, you flip through the whitepaper; then you check the team background; and in the end you still have to dig through on-chain data. But honestly, these days there aren’t many projects that can keep me up researching until 2 a.m. @NewtonProtocol is one of them.
What drew me in is actually very simple. In the past, when we used AI trading tools, the biggest headache was the black-box operation—you didn’t know how the strategy worked under the hood, and developers had no authoritative way to prove their code was reliable. The result was that good strategies nobody dared to use, while bad projects kept cutting one batch after another through marketing. What NEWT wants to do is to install a layer of trust infrastructure for this chaotic track. It aims to build a secure aggregation layer where AI strategies, trading bots, and the developer ecosystem run transparently on-chain—code is verifiable, performance is traceable, and anyone with real skill can be recognized at a glance.
For someone like me who cares about technology and also about capital safety, this hits the pain point squarely. Imagine a future where every AI parameter you set and every backtest you run is stored and attested on-chain. Developers can attract users with real performance, and users can confidently hand over their funds to verified strategies. Isn’t that exactly the kind of trustworthy AI trading market everyone has been hoping for?
Of course, the ideal is full and beautiful—the reality still needs to be proven in deployment. Whether the team can build robust security and verification mechanisms, and whether the developer ecosystem can really take off, will all require time and validation. But in 2026, when AI and Crypto accelerate their integration, the direction and entry point NEWT has chosen has already, at least for me, earned a spot that I can keep tracking.

@NewtonProtocol #Newt $NEWT

June 30, #alpha 6 — Alpha Airdrop preview! 11.5W participants!
📅 Today’s airdrop
At 18:00, in 224 minutes, there will be a blind box airdrop ambush. It’s time to test your luck—may everyone pull the big prize 😁
Yesterday, the total trading volume from limit orders reached 1.66 billion U, a 7.7% increase.
Binance Alpha Trading Competition report:
1️⃣ ARX is still the big boss! 0.2614, traded $14.68M in 24 hours, up 3.59%, with total volume at 4.138B.
2️⃣ KGEN is chasing a real culprit! 0.1988, nearly 900k trades completed, total volume 645M, with 3 days left. But the gap vs ARX is still big—whether it can make a comeback depends on whether big funds enter next.
3️⃣ STABLE must be singled out for praise today! Up over 8.56%—a black horse among black horses! 24H成交 8.16M, total volume 1.155B.
4️⃣ STAR 0.1271, down slightly by 3%. Total volume is 283M—its base is already there. Once it washes out, it may soar again. Keep an eye on it.
5️⃣ GWEI 0.1578, down 2.11%, total volume 355M. Although it has retraced a bit, it’s still holding onto the top five—resilience is good.
Today’s trading recommendations (new coins within 30 days, points ×4)
Recommend ARX—22 days left. Do trades around 200–500 U per entry, and refresh in small batches.

#opg $OPG When researching projects that combine AI with blockchain, I’ve found a common pain point: storing large model data on-chain is too expensive. Only after seeing how @OpenGradient handles it did I feel there’s a more reliable solution.
It doesn’t shove those huge model files—often tens of GB—directly onto the chain. Instead, it sends everything to Walrus decentralized storage. The approach is very practical: package the model files, inference proofs, and all those big components into Walrus, and on the main chain it only records a Blob ID as a pickup code. When nodes need it, they use that ID to retrieve the data from the repository. The mainnet ledger stays lightweight and won’t be overwhelmed by massive data.
In the past, some projects insisted on putting everything on-chain, which led to network congestion and Gas costs that were wildly high. OpenGradient’s design—separating storage from accounting—fully decouples bookkeeping and storage, so the burden is indeed much lighter, and the workflow runs more smoothly.
That said, we should also mention the role of $OPG . It’s the core link of the entire ecosystem. Whether it’s occupying storage space or nodes downloading models and providing compute power, everything is settled using OPG. This creates real incentives for those providing hard drives and compute resources—this isn’t just talk about decentralization; it’s genuine, tangible benefit alignment.
From the documentation, it looks quite solid in terms of execution. The HACA hybrid verification path also balances speed and verifiability. Different modes—TEE, ZKML, Vanilla—each have their focus, giving developers flexibility in how they implement things. Of course, AI on-chain is still in the early stage, and performance/cost compatibility still needs exploration. But the practical “reduce the load + incentivize properly” direction shown by OpenGradient is definitely worth a closer look.
Oh, and in July there’s another OPG unlock. The project team’s proportion of token holdings isn’t low, and the ecosystem is continuing to advance. If you’re interested, you can check their Model Hub and technical docs and even try running it yourself. $OPG doesn’t play games—it uses real storage and settlement to address the “memory anxiety” at the intersection of blockchain + AI. This idea might be able to go even further.

On August 29th, #ALPHA Alpha airdrop announcement! 11.2W people!
📅 Today’s airdrop
No airdrop announcements yet.
Yesterday’s limit-order total trading volume was 154.5 million U, up slightly by 0.06% from the day before—basically range-bound, not much movement.
24H Trading Competition data:
ARX took the leaderboard, doing 13.39 million U in 24 hours, and pushing total volume to 29.7 billion—steady as an old pro. Even though the price pulled back 7.88%, brothers really don’t care; spot and contracts both push, and volume tells the whole story.
Second is KGEN: total volume 36.9 billion, price 0.1931, down 14 points, but the volume is still rock-solid—the whales really do have deep pockets.
STABLE is third: total volume also 36.9 billion, going face-to-face with KGEN in terms of volume, price 0.036, FDV 3.6 billion U.
The hottest today is GWEI: up +23.36% over 24 hours, one of only two green charts in the whole room. It surged straight into 4th place on the overall board, with trading volume hitting 297 million.
Fifth is STAR: total volume 246 million, up 1.45%—steady with a little seasoning.
Today’s recommendations (new coins within 30 days, points ×4):
Pure-volume pick ARX (23 days). Do small multi-lot entries, roughly 200–500 U per trade, then cycle back and forth.

Just finished scrolling through messages from a few groups, and I found that $OPG has become a hot topic again. Some people are saying “buy the dip,” while others are cursing “it’s distribution/exit.” The arguments are going back and forth endlessly. I, on the other hand, feel pretty calm, because I’ve been tracking this project for quite a while—I know in my heart what it’s good and bad at.
@OpenGradient ’s vision really resonates with me. Having intelligent contracts use a model to reason—that’s no longer simple if-else logic; it gives on-chain applications real intelligence. DeFi protocols can monitor the market, rebalance, and mitigate risks on their own, without relying on off-chain bot arbitrage. That way, overall capital efficiency can take a step up. The fact that a16z and Coinbase Ventures are willing to put money in shows this isn’t an off-the-cuff idea. The on-chain data is also solid: more than 2.0 million verifiable inferences have already run through—it’s not just a “whitepaper hype” stage.
What truly made me decide to hold my position, though, is how they restructured the way they charge for compute. Ethereum’s Gas model is too coarse-grained—running a simple transfer shouldn’t cost the same as running a big-model inference. OpenGradient brings the inference’s topological complexity, the model’s parameter count, and the validation mechanism into the pricing, forming a dynamic, perception-aware pricing system. If this mechanism can run stably, $OPG ’s value has an anchor. It becomes the real unit of measure for compute consumption across the network— the more prosperous the ecosystem is, the more valuable it is.
Now the market is really grinding people down. The weekly RSI is still churning in low territory, the MACD green histogram can’t shrink, and volume looks lively, but the price simply won’t move up. The selling pressure from token unlocks is like a dull blade slowly cutting you down. At a time like this, adding a big position doesn’t seem very rational. I’ll just hold the same amount I had before and keep observing—waiting for the RSI to move out of extreme levels or for the MACD to show signs of narrowing, then I’ll judge whether this is truly the bottom.
The combination of AI and blockchain is definitely a long-term direction. But whether OpenGradient can make it through this winter depends on the subsequent product rollout. There aren’t many projects bold enough to break old rules—I’m willing to give it some patience. The market’s answer will come with time. #OPG
What do you think about OPG’s next move?

Last week while browsing GitHub, I noticed contract call data for $OPG : the testnet’s average daily request volume has quietly climbed to nearly twenty thousand. That number isn’t shocking in the AI encryption race, but compared with three months ago, it has grown almost sevenfold. What @OpenGradient wants to do is the foundational layer for calling AI capabilities. The positioning sounds grand, but the reality is harsh—without real developers pulling data to run inference, the token economy is just an air castle. Right now, $OPG is stuck in a phase where the story is polished but validation hasn’t caught up.

I’ve personally had firsthand experience recently while leading my team to build cross-chain DeFi tooling. Two months ago, I wanted to integrate an AI smart routing layer into a yield aggregator. In the end, we discovered we had to rewrite contracts, migrate liquidity, and re-train users. After three weeks, progress was nearly zero. I used to think EVM compatibility was a constraint for AI—given how different the compute requirements are, we should start fresh. But when we actually did it, we found users don’t want to move, liquidity gets fragmented, and AI ends up lacking data support. Looking back, enhancing on EVM may start slower, but we can directly reuse existing TVL and tooling—so the iteration pace is actually more stable.

OpenGradient is exactly this approach. On EVM-compatible networks, developers can call AI inference via precompiles, with almost no need to change Solidity code. We’ve been running on the testnet for two months: the aggregator passes in positions, price spreads, and market sentiment. With a single call, you can get the best recommendations backed by TEE proofs—the output is verifiable and can directly be used for conditional logic inside smart contracts. Cross-chain data can also be unified and fetched: Base, Arbitrum, and Optimism no longer each operate on their own. Our team has already fully integrated it, and the efficiency gains are genuinely noticeable. If you’ve been dragged through the various pitfalls of AI integration in the EVM ecosystem, this direction is worth spending time researching.

Of course, smooth technology is only the first step. Modulus and Giza are watching this space too. Ultimately, who remains will depend on real paid conversion after mainnet launch. Whether incentive-driven data can solidify into network effects—no one can guarantee that.

As for me, I’ll keep staring at the testnet data. The day average daily requests break 100,000 and the share of paid calls exceeds 30%, then $OPG will truly be “running.” #OPG

#ALPHA 6June 27, Binance Alpha 30-day newly listed token trading competition
ARX current price 0.2866U, 24h trading volume over 14 million, up a little more than 3%, FDV 286 million. Today the limit order volume reached 450 million, yesterday it was only 86 million, clearly there is a big buyer supporting the orders underneath.
CAP current price 0.0294U, trading volume directly reached 71 million, soaring 194%! FDV 294 million. Today the limit order volume is 150 million, yesterday it was 16 million, and the money is rushing in like crazy. Note: there are still 29 days until expiration, a typical short-term casino. If you're slow, don't chase; if you've entered, set stop-losses, and don't slap your thigh over a pullback in profits.
O current price 0.4238U, trading volume 10.4 million, down 8%. FDV 423 million, 20 days until expiration. This pullback is pretty healthy. If you want to buy the dip, you can place staggered orders; the fundamentals aren't broken, it just takes patience.
NES current price 0.1794U, volume 11.5 million, down 12%. FDV 179 million, 27 days until expiration. The short-term action is a bit miserable, but it hasn't broken support. Those daring to bet on a rebound can try with a small position, don't go heavy.
H current price 0.0589U, trading volume only 210,000, up 3.7%, FDV 589 million. The volume is the smallest, 20 days until expiration.

A warning: these coins only have about 20 days until expiration, all of them are short-term sentiment plays, coming fast and going fast too. Manage your position size well, don't go all-in on impulse. Take profits and run is never wrong; don't expect to sell at the absolute top. Brothers who chase higher prices, take it easy. Better to miss than to make a mistake.

#alpha 6月27号，Alpha airdrop announcement! 10.8W participants!
📅 Today’s Airdrop
There is no airdrop today. Eat, play—enjoy your day.
Today’s total limit order transaction volume is 1.53 billion, up 0.06% from yesterday. Basically it’s range-bound with no big movement.
Binance Alpha 24H Trading Contest:
#1 ARX: 24-hour trading volume 82.87 million, price holding steady at 0.2840, up 2.13%, with FDV around 284 million. The most impressive is yesterday’s limit order成交 amount of 43.6 million+—who wouldn’t find that data confusing? Today’s heat is absolutely in the C position.
#2 KGEN: trading volume 47.89 million, up 1.84%, current price 0.1898—quite steady.
#3 STABLE: took off today, up 4.37%, with trading volume of 17.41 million.
#4 GWEI: up 4.32%, trading volume 23.78 million—steady, no-nonsense type.
#5 STAR: pulled back a bit today, down 5.57%, but trading volume is still over 16 million.
Today’s recommendation (tokens to be launched within 30 days, points ×4):
Pure trading volume picks: ARX (25 days left). Do 200–500U per trade—small amounts, multiple times.

At the end of last year, I had some free time, so I put together a list of all the on-chain AI projects that could actually run in the market and tried them one by one. The result? My wallet got noticeably thinner, my GPU was almost smoking, and only a few were truly worth it. Some projects had websites that looked like top-tier Silicon Valley unicorns, but as soon as you called the API, it timed out; others claimed to do decentralized inference, but ran slower than my grandmother browsing the internet. Not to mention all the annoying stuff like random disconnects, lost context, and mismatched results.@OpenGradient
Then I came across OpenGradient Chat. To be honest, I didn’t have high expectations at first—I figured it was probably just another rebranded pitch deck. But after using it for two weeks, I was convinced.
It should be one of the earliest on-chain conversation tools adapted for Claude Fable 5. Its distributed nodes use layered encryption with hashed salt values. I often use it for batch data analysis, sometimes for half a day at a stretch, and the model context stays rock-solid with basically no interruptions. I honestly didn’t expect an on-chain project to achieve this level of stability.
What impressed me most was its private chat section. It uses the uncensored Nous Hermes model, separately isolated in a privacy-chain session, and the underlying protection is quite solid. My current habit is to use the public area for reviewing market trends and organizing materials with Claude Fable 5, while the private area is dedicated to digging into logic in depth. The two sides do their own thing without interfering with each other. And holding $OPG can unlock priority compute; the public and private compute channels run independently, so there’s no fighting over resources. Those so-called dual-model setups from other platforms are, to be honest, mostly just marketing gimmicks—the underlying systems share the same pool, so when things get busy they slow each other down.
Technically, it really has some interesting ideas. The HACA design separates execution from verification: the GPU focuses on inference, while the Full Node only verifies TEE attestations or ZKML proofs, so every node doesn’t have to recompute everything. In this way, trust is split into different cost tiers: TEE for speed, ZKML for certainty, and Vanilla for a lightweight path. In essence, verification authority has been made optional, which is pretty clever.
There’s also PIPE, which moves computation earlier in the flow so results come out first and only then do you decide whether the transaction is valid. That approach is pretty rare in on-chain AI. MemSync tries to connect the context across different models and applications, and that has a lot of potential.
Of course, to be blunt, strong technology doesn’t necessarily mean the ecosystem will take off.#OPG

#alpha 6月26, Alpha Airdrop preview!
📅 Today’s airdrop
Binance Wallet CAP new listing time has changed to 18:00–20:00. Deposit 3 BNB, 225 minutes, and once the round ends, you can sell right away. Follow the pre-market price trend for reference. This airdrop’s total amount is roughly around 2.25 million. Set your alarms accordingly.

Yesterday, Alpha’s limit orders had a total traded volume of $1.521 billion, which is down by just over 13% compared to the day before. Hype cooled slightly, but overall it’s still playable.

Trading contest:
- PRL: Yesterday 175,691; today surged to 194,916—up 19,225 places!
- ARX: Yesterday 38,699; today went up to 69,996—up a massive 31,297 places!

Today’s recommendations (tokens launching within 30 days, points ×4):
Pure trading volume pick: QAIT (only 1 day left). 200–500U per order, and make multiple small orders to refresh momentum.

June 25 Alpha Airdrop Announcement! 11.2W people!
📅 Today’s Airdrop
Tonight 18:00–20:00, Binance Wallet launches CAP subscription for the new listing. It takes 3 BNB—key point: once the activity ends, you can sell immediately; no need to lock your funds. Blind guess: it’s a high score!
Besides the CAP subscription, don’t forget the OPG creator event in the Square as well.
Lately I’ve been getting a bit numb watching AI projects on-chain—most are basically wrapped in RAG and claim to be decentralized intelligent. But OpenGradient Chat surprised me a bit; it really runs the model on-chain.
I tried its on-chain quote—the whole process was pretty interesting.
After sending a request, the x402 fee was charged almost instantly. Then the inference result came out in the next few hundred milliseconds. Finally, that verification proof landed slowly at the end. I tried several times—the order is always the same: money goes first, the result comes in the middle, and the proof is filled in at the end. The execution nodes are busy doing inference, while the verification nodes generate the proof—two paths running asynchronously.
The project is actually quite generous. It supports TEE hardware isolation, ZKML mathematical proofs, and a baseline solution—but the path selection and risk assessment are all handed to me to decide.
But the issue is exactly here: when I use the result, I have no idea which step the verification has reached. The whitepaper says it can be checked afterward, but after—how long? There’s no clear window given. This isn’t a technical detail; it’s a trust delay. I’ve already used the result, but the verification is still on its way, and I don’t have peace of mind.
That said, the product idea is quite solid. Put the conversation into a TEE to run it—externals can’t see the content—and then package the hash and put it on-chain. @OpenGradient pushes privacy and verifiability one step forward. If stability keeps up in the future, these scenarios—DeFi automation, strategy execution—really do look promising.
Of course, the technical implementation is still early. We’ll need to watch the on-chain data over time: can it handle high concurrency? how are node incentives designed? are there backdoors in the hardware?
Also, $OPG —after getting listed on Binance and other major exchanges, liquidity improved, but the price didn’t hold up. Listing on big exchanges was supposed to be a positive, but in the short term it looks more like opening a door for liquidity to be the one taking the bag.
#OPG

#alpha 6, June 25th, Alpha Airdrop Alert! 109K Participants!
📅 Today's Airdrop
Currently sitting at zero, but tomorrow's got a big one! June 26th (CAP) from 19:00 to 21:00, new Binance wallet launch, 3 BNB.
CAP's tokenomics show an initial circulation of only 15.6%:
- Market Makers 0.6%
- Public Sale 5% (those who bought in at a $110M FDV, fully unlocked on TGE)
- Ecosystem 10% (includes 1% from Binance’s new launch and 0.2% from Boost)
The key point is, the community isn’t issuing new tokens this time, only stablecoins! This is roughly equivalent to 5% of CAP targeting a $250M FDV, about $12.5M. So besides the public sale, Alpha, and Boost portions, the rest of the tokens are tightly held by the project team. Those in the know will get it, keep your heads up.
2. Yesterday’s Limit Order Execution
Total limit order trading volume hit $1.75B yesterday, up 3.28% from the day before—steady but at least it's climbing.
3. Trading Competitions
- BILL trading competition ends tonight at 21:00! Ranked 278,231 yesterday, now up to 286,476 today, jumped 8,245 spots in a day!
- PRL: from 165,597 yesterday to 175,691 today, up 10,094 spots, solid gains!
- ARX: this one's a beast, from 5,039 yesterday to 38,699 today, skyrocketed 33,660 spots, today's champion!
4. Today's Recommendations (Tokens launched in the last 30 days, points ×4)
Based purely on trading volume, recommending QAIT (2 days), trades between $200-$500 each, small amounts to grind.