Native News and Research for Crypto | AI | Macro Enjoy: - Our Article Highlights - Daily Outlooks 📊 - Weekly Scope Reports 🔎
Live Updates ⧭
- Tweets 34,032
- Following 480
- Followers 3,251
- Likes 60,075
Nick Alman on Solana, Jito & the Future of Crypto
“We are in, I think, about third-wave crypto. We’re kind of moving into now.”
“Fundamentals era is coming. We’re starting to have to build real things.”
“The narrative in Solana that has been emerging over the last few years—we call it internet capital markets.”
“We are trying to build a kind of layer 1 blockchain that effectively can compete with something like the NASDAQ.”
“For that to happen, we need extremely low latency. We need complete robustness. It needs to be up all the time.”
“The blocks are getting faster. Bandwidth is increasing. The amount of compute we can do in every block is increasing dramatically.”
“We’re just trying to create the kind of substrate that allows all of the capital markets in the world, all of them, all of the real-world assets, not just crypto assets, to come on-chain.”
“Be the kind of 24/7 trading environment and the place where you can raise money and build products and be this kind of, like, everything chain for the frontier of finance basically.”
1
70
i'd add one more check that has nothing to do with price
is the thing you're betting on still growing
usage, fees, devs shipping, whatever the thesis actually rests on
if those are fine while everything bleeds, it's just beta
if those are rolling over too, "more time" gets a lot harder to justify
type of day where my core coins are down
one of those things where as a gut check to see if your thesis is wrong at its core, or if it's a timing/correlation thing:
- what does BTC look like?
down and looking shaky on the weekly, people doing hand-wringing about the latest breakout being an SFP
- what do other coins look like?
coins that i hold on other chains/platforms/ecosystems are also sharply lower the past few days
- what do coins that are competitors, or diametric opposites look like?
also in a downtrend the past week
--
so in short, it's less "my thesis looks incorrect fundamentally", and more "we're in the early bull market volatility and illiquidity zone", where clearly there are a lot of bears still GOONING about any prospect of BTC accepting back into a lower range, or below an MA, and those bears know they're on a shot clock to try to make something happen
if bags central to a thesis were down in isolation, ie:
BTC is pumping + grinding up (rinse and repeat)
other ecos are having broad, multi-day/week pumps
coins on the opposite side of this thesis are enjoying strong/sustained inflows,
that's where you ask yourself some hard questions, and consider picking up the ball and going home
right now, the answer seems to be "more time"
1
41
Qwen3.8-Max has been sitting at the top of the Chinese models on Artificial Analysis and almost nobody is talking about it.
The versions that actually spread are the smaller ones. Qwen3.8-27B and Qwen3.6-35B-A3B keep getting forked, quantized, and shipped into local stacks.
The community still routes around the biggest Qwen when a mid-size checkpoint is good enough and easier to run.
32
i'd add one guard: don't open a new pr if the last one isn't merged yet
hourly small prs sound great until 12 are sitting in review
also worth letting it flag when a "violation" shows up everywhere
sometimes that's the codebase telling you the standard is wrong
Thinking about making a skill called /fix-one-thing:
"Read CODING_STANDARDS.md. Find a violation in the codebase, and fix it. Make the PR small and easy to review, with a small blast radius."
Run it whenever you've got a spare moment, or once an hour on a schedule.
2
2
2,986
The hardest part of a massive crypto trade is usually the time between entry and exit.
Finding the narrative early is one skill.
Holding through 70% drawdowns, months of boredom and nonstop FUD requires a completely different one.
Being early gets you into the trade.
Having the discipline to stay there is what captures the move.
normal ppl dont understand the extreme level of neurodivergence required to hit these massive trades
u must have extreme adhd to be early to a narrative + main coin
then, u must have the exact opposite, extreme patience/discipline, to be able to hold that coin through 70-90% drawdowns, periods of zero volatility, and endless waves of fud
one mindset to enter, complete opposite mindset to exit
few
40
CZ on the FTX Collapse
“As part of Binance’s exit from FTX equity last year, Binance received roughly $2.1 billion US equivalent in cash, BUSD and FTT.”
“Due to recent revelations that have come to light, we have decided to liquidate any remaining FTT on our books.”
“Sam Bankman-Fried, the co-founder and CEO of FTX, says your tweet killed FTX. Is there any part of you that thinks he’s right?”
“No, absolutely not.”
“If one of your competitors can make a tweet to kill your company, then you don’t have a company.”
“If he was running a company that I could kill with one tweet, he would never have that company.”
“I could tweet about any company. They will not go down.”
“The fact that they went down, there were multiple reasons.”
“No company goes down for one tweet from a competitor, no matter how large that competitor is.”
1
61
GPTZero 4o takes first on this third-party detection board.
It leads Pangram 4 across most slices, including multi-domain, multi-LLM, and LLM-generated text.
The dip is the same for both. Time-test and human writing still break detectors more than raw generation does.
Detection is getting better on obvious AI text. The hard cases have not moved as much.
35
the doc used to be proof someone actually thought the design through
writing it was the forcing function
now you get the artifact without the thinking behind it
and reviewers end up asking ai to summarize something ai wrote lol
I'm not entirely sure how to navigate the trend of technical design docs being vibed.
The information density, the odd writing tone, and the standardized presentation style across all the vibed docs generated from the same model are making technical artifacts human un-readable
52
90/10 only sounds fun in hindsight, in the moment nobody knows which 10% window they're in until it's already closing
Crypto is a miserable market 90% of the time
And it's absolutely godlike 10% of the time
Knowing this why the fuck are so many of you trying so hard to be miserable 100% of the time
Just enjoy when this shit pumps
It won't be good forever
It's simply a brief window of bliss
42
Nate Soares estimates Claude Mythos became one of only three entities that could take over your phone just by getting you to open a website
"One of the holy grails of hacking is can you make a website where if you just look at the website, I get full control over your computer or your phone."
"I don't know the exact numbers because I don't have the security clearance to know the exact numbers, but a decent guess is that in January of this year, there were only two entities that could pull that off. Mossad and the NSA."
"In March of this year, there were three. Mossad, the NSA, and Claude Mythos, which is this new AI made by Anthropic."
1
51
OpenCode's sitemap leaked the next wave before any lab posted it.
GLM-5.5 Flash. GLM-5.4. Kimi K4. DeepSeek V4.1 Pro. Muse Spark 1.4.
They pulled the listings. The pages are still live. Kimi K4 already has a stats page sitting at 0 tokens.
The best leak source in 2026 is an SEO file.
123
Almost 48 hours on GPT-6 Sol and Claude Opus 5.5.
Sol on High, $200 plan, no Fast Mode. Full day of heavy use, about 8% of the weekly limit.
Opus 5.5 on Mid, $20 plan. Same 8% used, and 4 to 5 client site drafts out of it.
The limits finally feel usable. The better move is not picking a winner.
One builds. The other challenges. That pairing is the real stack.
60
Opus 5.5 shipped with new measures to stop anyone extracting Claude's reasoning.
The same week, Xiaomi published MiMo-V2.6-Pro's RL code and training environments under MIT.
Anthropic is guarding how its model reasons.
Xiaomi is giving away how its model was trained.
Anyone training a reasoning model can start from Xiaomi's setup.
40
Top 6 gainers and losers over the past 7 days
Gainers:
$NEON: +708.9%
ethereum:0xaaaaaa20d9e0e2461697782ef11675f668207961: +412.9%
solana:8SMMso8Muv8d6i4WmMDthKt6TN1ysN6937sx3DKLXZqB: +343.0%
$ETN: +219.9%
saga-2:native: +202.6%
ethereum:0x7cf9a80db3b29ee8efe3710aadb7b95270572d47: +173.3%
Losers:
solana:CWZ6BsdnjkDVTGkmL6bGbJXXig6ceef12KvyGQW14cMt: -85.0%
solana:HFsC98RB5jb2AsouJEt3Pv7baC2szbg8PQhD2Xwjpump: -55.3%
solana:HcRLc9VDgjLeK154xDawfb1dmVJ98DoSqcwTHGqiDeJR: -50.1%
$LAPTOP: -45.3%
$GRX: -38.3%
$AI: -37.7%
141
Top 10 chains by active RWA market cap
About $36.6B sits on the top ten. For the first time, the other nine together exceed Ethereum.
• Ethereum – $16.6B
• BNB Chain – $5.78B
• Solana – $4.49B
• Stellar – $3.36B
• Avalanche – $1.74B
• Liquid – $1.61B
• Arbitrum – $1.05B
• ZKsync – $971M
• Polygon – $509M
• XRP – $456M
Ethereum is still first, but its share is ~45% versus ~48% last week after slipping from $17.3B to $16.6B. Issuers have not left; new issuance is landing on other chains.
BNB Chain is the main gainer, with bStocks tokenized equities lifting it clear of Solana. Stellar and Liquid matter more than their DeFi footprints suggest as both hold billions in funds and securities.
In RWA, issuer venue still outweighs on-chain DeFi activity.
48
US crypto enforcement peaked in 2023, when the SEC drove most quarterly actions. Since 2024 the count has fallen, and 2026’s pickup is almost entirely CFTC, not SEC.
2
3
51
GPT-6 Astra uses 27k output tokens per Intelligence Index task.
Claude Opus 5.5 uses 119k, the most of any model listed.
GPT-6 Luna, OpenAI's cheapest tier, uses 51k, nearly double Astra.
On subscription plans, Opus 5.5 burns through usage limits more than 4x as fast per task as Astra.
1
52
Prediction markets, ranked by who actually has a book.
S tier
@Polymarket — still the only one with real depth.
@Kalshi — biggest book in the category. Fiat, regulated, sports-heavy. Not onchain.
@OG_com — Crypto(.)com’s regulated US book. Real volume, fiat rails.
@RobinhoodApp — Kalshi/OG distribution, not its own book.
A tier
@predictdotfun — strongest BNB book.
@opinionlabsxyz — BNB macro/crypto challenger that actually has flow.
@trylimitless — best short-dated crypto markets on Base.
@azuroprotocol — infra, not a front-end. Powers a lot of sports apps.
@HyperliquidX — native markets for people already on HL.
B tier
@MyriadMarkets — media-embedded, casual size only.
@DriftProtocol — Solana, capital-efficient, thinner books.
@SX_Bet — sports-first onchain venue.
@Overtime_io — onchain sportsbook. Deeper props than Poly, smaller size.
@Novig — US sports P2P, regulated. Not crypto-native.
C tier
@GlimpseMarkets — Bitcoin-native range book, focused on price forecasts, not politics.
@xomarket — permissionless markets, liquidity still split.
@Predictstreet — FIFA-tied sports book, not a crypto-native venue.
@magicmarkets — sports aggregator, thin vs the majors.
@TurboFlow_xyz — short-dated event contracts, still small.
@AugurProject — OG protocol in reboot mode. Almost no live book.
@PancakeSwap — short BTC up/down.
@PredictIt — old US politics cap. Tiny vs Kalshi/Poly.
D tier
@manifold_markets — play money. Good for practice, not real size.
@metaculus — forecasting site, not a live crypto book.
@hedgehog_markets — Solana leftover, barely moving.
@Polkamarkets — old Polkadot experiment. Dead volume.
1
78