“Abstract, we are closing down.” I’m not even surprised. The next shutdown might be shocking We already know who’s shutting down next this month?

just hopped on @duel_duck to see what people are cooking and found this one: will BTC hit $90k before October 8, 11am UTC? normally i’d let the market do it’s thing, but i’m taking NO on this one BTC is sitting around $83.7k right now that’s roughly a 7.5% pump in less than a day possible? sure, but realistically im not seeing it around that figure soon so it’s an easy NO for me what are you taking, YES or NO?

been reading the @BaranosAI whitepaper, which is built on Fogo and i'm especially interested in the milestone-gated DAO grants use case. today, deciding whether a grantee actually hit a milestone usually comes down to a multisig or a vote. the problem is that this often happens after the work is done, when everyone already knows who benefits from the decision. that's where politics can creep in. baranos' design points to a different setup. before the grant even starts, the DAO commits to the model, rubric and evidence rules as an inference policy. the grantee then submits evidence — a repo, audit, onchain activity, etc. — which gets hashed and frozen at the deadline. an executor runs the inference offchain and posts the result onchain. if nobody challenges it, it settles. if someone does, the disputed computation can be replayed onchain. then the DAO's treasury contract simply reads the settled result and decides what happens next. baranos never touches the funds. and honestly, this is the part i find most interesting: the interpretation gets locked in before anyone has a stake in the outcome. the paper puts it well: the goal isn't to remove interpretation, but to make the interpreter, evidence and rules explicit beforehand. there are still caveats. verification proves that the model ran according to the committed policy. it doesn't prove that the model itself is correct. and the paper still needs public benchmarks around latency and cost. also, the initial focus is prediction markets, so milestone-gated DAO grants are more of a proposed extension than the core use case. but if this architecture can work as intended, i'd take a judge whose rules were fixed on day one over a judge deciding the rules on day 90.

AI inference costs can add up fast. I looked at DeepSeek V4.1 Flash as an example of how pricing can differ between platforms. @say_gm_ currently lists the model at: $0.195 / 1M input tokens $0.78 / 1M output tokens That’s 35% below DeepSeek’s published list price of $0.30 input and $1.20 output. The bigger point is: when you're running AI workloads at scale, lower inference pricing can make a real difference to your overall bill. Same model fr, Lower listed pricing through SayGM. Always check the live comparison for current rates, since pricing can change. Now sayGm😎💛.

Something I bet you didn’t know about AI Every time you use AI for complex tasks, you pay for two things Input: your prompt Output: what it writes back I’ve compared the cost prices of GPT-5.6 Luna on both platforms For open router Input : $0.21 Output: $1.27 For saygm Input: $0.17 Output: $1.01 It about 20% cheaper on @say_gm_ If you’re an AI MAXI and you run automation and coding all day,those savings adds up fast Same model, Lower price with @say_gm_

A chance for you to make money, even if you have a small account. I know how most creator programs work. By the time they start getting attention, the room is already packed with creators. But @1win is still early. Fewer than 30 creators have been approved so far, which means there’s still an opportunity to get in before things get crowded. Here’s what they’re offering: • $1,000+ USDT weekly prize pool for creators • Special Discord role + access to private channels after approval • Ambassador opportunities for top-performing creators I just joined the program myself, and I think it’s worth checking out if you’re already creating content on X. Apply here: 1w-ugc.pro⁠ Registration code: CMU6ARH Good luck 🫡

i tried opening a trade on @Hertzflow_xyz and got stuck I'm so used to seeing bids, asks and an order book before opening a perp that I was literally looking for something that isn't there so i started figuring out how HertzFlow actually executes trades. and that's when I started noticing what was different #Hertzflow 🧵

just yesterday I was lamenting on the TL about my claude sub expiration and I've been torn ever since :( since I use a lot of Ai credits daily especially with Fable 5.1, so I went hunting for savings and found @say_gm_ Claude Opus 5, per 1M tokens (input/output), from SayGM's comparison page: • SayGM: $4.47 / $22.33 • OpenRouter: $5.28 / $26.38 OpenRouter's figures include its 5.5% credit-purchase fee, so this isn't purely a token-price gap. Estimate for 10M input + 2M output tokens a month: $89.36 vs $105.56. About $16 less, roughly 15%. Why it matters: small gaps add up across coding sessions and automations. When you can't afford both tools, that's real money. SayGM says top-ups are credited in full. I haven't run my own workload yet, so this is published pricing, not savings I've measured. Prices as checked Oct 7, 2026. Savings vary by model. get started > saygm.com/vs/openrouter

a creator with 3k followers can bring more value than someone with 500k. sounds obvious, but most creator campaigns still don't really work that way. reach usually wins. that's one thing that caught my attention while digging into how @0xhazels approaches creator reputation. the idea isn't that having a huge account automatically makes you more valuable. originality, content quality and the conversations you're actually able to create matter too. and I think that's important. because someone with a smaller audience can drop one genuinely good piece of content that gets people talking, thinking and discovering something new. while someone 100x bigger can post something lazy and still farm numbers because they already have distribution. those two shouldn't automatically be valued the same. attention matters. but the quality of that attention matters too. that's probably one of the more interesting things I've found while exploring Hazels so far.

You’ve probably seen a dozen “massive token burn” announcements by now, and some of them aren’t really burns at all the tokens just move to a wallet everyone promises not to touch, and they still count toward supply @streamflow_fi just burned around 70% of the original $STREAM supply, and this one is different the tokens are actually gone forever , supply went from 1B to 300M, and nothing can be minted back here’s what happened and why it matters Let’s take a tour 🧵

the monster said "grrr" and then gm like it was a normal greeting... and then threatened us with lil virus dna infection too. @kairopeng, who is supervising this lab

good opportunity right here if you’ve got some stablecoins sitting around, you can actually put them to work instead of leaving them idle @Canborsa_DEX lets you deposit and stake USDT, USDC and USDCx while earning up to 20% APR and while your stablecoins are earning, you’re also collecting points based on how much you have staked you get 0.5 points per $1 every day, so staking $1,000 would give you 500 points daily you can also withdraw whenever you want, so your funds aren’t locked in the pool is capped at $300k though, so there’s a limit to how much can go in join here ↓ canborsa.com/r/5YACPKJN btw, here’s a simple guide on how to deposit and stake

$100 for 2 posts 🎉🎉 I'll accept it, bring it please.

agents can already find what you want to buy but paying is still where it stops. so i was researching some days ago and randomly came across @agentcardhq. apparently they're backed by @ycombinator and the idea is simple. give the agent a card. say an agent needs to order something online, book a ride, or pay for food. it can get there, however, checkout still needs you. Ideally, you don't want to hand over your real card, and a lot of sites still block these agents. that's where they come in you can save the card you already use and approve the payment on your phone or they make a new card, with a spend limit. it can be one-time or reusable. also, the money sits in usdc on base, so it spends from a limit, not your whole balance. they're still pretty early and mostly useful if you're building something that buys for users. you can check for more details on their website in the comments. disclaimer: this isn't a paid post. I just like what they're building and felt like sharing.

lazy day