Eduardo Moreno

@unusual_whales Hoooooly shit. Meta really enjoys being sued to oblivion apparently. Glad I didn’t stay there for very long to see the 2026 shitshow.

Meta, $META, reportedly used AI to target workers with medical conditions for layoffs, per Reuters

It appears to only work with ChatGPT Business accounts, not ChatGPT Plus or Pro. Or at least, it forces me to sign up for ChatGPT Business from the link, despite being logged into my Pro account. That makes some sense, but you might want to clarify that, seeing as Plus and Pro also have usage credit pools available for when you hit the plan limits. Especially since the business plans have a minimum number of seats.

@afestekjian @every codex credits should definitely work with any existing subscription! im not 100% sure about the core product credits in posthog, will check and get back to you!

BREAKING: Introducing All Access from @every, our new membership tier for the best builders in AI All Access subs get the Builder Pack which includes $7,000 in credits and free usage to the models + tool stack we use @every. All Access subscribers get: - $1,000 in Codex / @ChatGPTapp for Work credits - 12 months free of @Cursor_AI Pro+ - $4,000 in @PostHog credits including self-driving to automatically fix bugs and identify issues in your production app - 1 year free of @Framer - 6 months free of @NotionHQ And much more! (Did I mention $1,000 in Codex credits? It's time to build!) Get all access: every.to/builder-pack Why All Access and the Builder Pack This is the best time in history to build something. For a long time, it’s been possible to one-shot impressive demos, but they’d fall flat the minute they hit production. But the release of GPT-5.6-Sol and Fable 5 heralds a new era: Everyone can build, launch, and maintain the software that they’ve always dreamed of. Everyone is a builder now. There’s just one catch: Building with AI is very expensive. (Ask me how I know.) (Alright, I’ll tell you. I accidentally used 2 billion tokens overnight this week on a big GPT-5.6-Sol run. Worth it.) This is unique in the history of technology. For most of the personal computing era, a billionaire and a solo builder could buy essentially the same top-of-the-line Mac. AI changes that: The more tokens you can afford, the more you can make. And we want to make that accessible to more people. That’s why the main feature of our new All Access plan is the Builder Pack: more than $7,000 in credits and discounts on the full stack we use to run Every, from idea to production—Codex, Claude, PostHog, Render, Gemini, FLORA, and more. Early-bird membership is only $500/year for the next 24 hours—and the Codex credits alone are worth $1,000. (I could’ve used it for my overnight run this week.) Now we’re handing it to you. Get all access: every.to/builder-pack Meet the Builder Pack It's got more than $7,000 in offers from 10 of the AI products we use to write, design, build, and run @Every: BUILD - $1,000 in Codex credits plus one month of ChatGPT for business - Twelve months free of Cursor Pro+ - One month free of @Claudeai Max - Three months free of @Google AI Pro DESIGN - One year free of @Framer Pro - One month free of @floraai Max HOST - $300 in @render credits IMPROVE - $4,000 in PostHog credits - Six months free of @Notion Business - Six months free of @AgentMail We rely on these every day, and we tried to put together a package that helps you comprehensively for each part of the process of building and running software in AI. What comes with All Access - Everything in an existing paid Every membership: our daily writing, guides, camps, and software like @usemonologue, @CoraComputer, @SparkleApp, and @TrySpiral - The Builder Pack, with more than $7,000 in partner offers - Unlimited email accounts use of Cora and unlimited Spiral usage - Members-only programming with me and the Every team and me Get All Access: every.to/builder-pack

It actually really has nothing to do with endorsement. Sanctions violations in the United States are treated completely differently and are a different crime than facilitating or aiding abetting criminal activity. Really sanctions are a means to exert pressure on a sanctioned party, be it a drug trafficker or an entire country like Iran, by depriving them of financial resources, typucally for reasons if national security. Anyone subject to the laws of whoever sanctioned them essentially has to: 1. not do business with them 2. in many cases freeze the funds of anyone who is sanctioned if they come into contact with them The whole point is to kind of essentially smoke the sanctioned party out of the activity that the sanctioning government doesn't want them to do so that they eventually acquiesce. In the case of the United States, since the US financial system is so deeply intertwined with the global financial system, being sanctioned by the US essentially amounts to a global constraint on your financial resources, Which is why people or countries or organizations sanctioned by the U.S. government have to turn to entirely underground financial systems in order to move money around because they essentially can't access, in most cases, global financial fund flows. So the whole reason why facilitating someone who is sanctioned financially is bad is not necessarily because you're endorsing them. It's because you are actively allowing them to sidestep national security policy. Now obviously receiving funds from a sanctioned party through a dusting attack isn't necessarily active facilitation but unfortunately the government has taken the position that receiving those funds and not acting per your legal obligations is still a legal violation.

@EdTheOSINTer @liamihorne Thanks for the response! If you don't mind a naive q: if the current onchain arch remains as-is, and corps *don't* file, wouldn't the legal system naturally figure out that receipt != endorsement?

New Tempo feature to prevent dust attacks! Also super useful for exchanges, onramps, etc - set whitelists on both tokens and addresses that are allowed to send to you.

For institutional recipients (and really anyone with sanctions obligations, which is technically everyone in a country with sanctions enforcement), this is a big deal. Prior to this, you had no choice but to receive whatever was sent to your address (acceptance is different from receipt, mind you). If those funds came from a sanctioned party (like someone on the OFAC SDN list, like how Tornado Cash was for a while), you actually have obligations to freeze the funds and file a report with the government. In institutions, this can be a huge compliance burden, especially since you have absolutely no recourse on chain but to receive the funds. Receive policies let you put a wall up in front of your recipient addresses so that deposits from addresses that are known bad actors (like sanctioned addresses) don't even enter your custodial perimeter. This is a huge deal for institutions operating onchain.

@liamihorne why is this worth the price (the authority)?

Receive policies are a unique feature on @tempo. On other public chains, anyone can send you any tokens they want. This can cause trouble in many ways: 1. A user could send the wrong token to a liquidation address (e.g., USDT0 to an address that is only meant to work with OUSD) 2. An address owned by a regulated entity could receive unwanted tokens (memecoins!) 3. An address owned by a regulated entity could receive tokens from unacceptable senders (non-KYC'd users, sanctioned entities).

@malleshpai @tempo As someone who has to actively deal with the situation on other chains where pre-custodial interdiction like this is *not* possible and you just have to accept whatever once broadcast, this is straight 🔥🔥🔥.

It actually seems to be getting worse. For some reason, once you get into more complex material or concepts, it often takes on this incredibly staccato-like prose that is incredibly difficult to parse. It's also not due to the terms being used (those are actually fairly uncomplicated), but the way it strings them together just makes it so unnecessarily obtuse, and then it behaves as if that's just how people write. It's like the models have to prove how smart they are, but it just forces me to have to reread the output more often than I ever would with human writing. The last paragraph in the retweet image is a decent example of that. The first paragraph is actually more like how I would want it to write all the time.

Anyone claiming AI writing isn't obvious anymore is a terrible writer I have noticed zero improvement between GPT5.5 and Sol when it comes to writing "humanity"

@0xShyam0x @mattpocockuk It feels kind of ironic that you say that while having this pinned to the top of your profile

@mattpocockuk yeah just like how bun's rust porting isn't idiomatic rust anymore with tons of unsafe keyword & guess what it works..

Yesterday I submitted a Claude plugin for my skills (finally) Anyone know how long the submission process takes? Would love a nudge from any of my mates at Anthropic

@mattpocockuk @towards_AI PR in this repo maybe? I see third party plugins in there: github.com/openai/plugins There’s also the ChatGPT plugin directory that was launched last week: learn.chatgpt.com/docs/submit-pl…

@towards_AI Where do you submit them?

BSA is still only a US law, and the most onerous KYC/AML requirements and stringent regulatory scrutiny come out of the EU. Honestly, in a risk-based program, your US-based users tend to be the least compliance burden, since traditional finance already has a mature compliance framework and KYCing users is dramatically easier thanks to external verification databases. The same can’t be said for RoW, particularly in Europe and LATAM.

@jayhinz The biggest GDP unlock would be reforming the bank secrecy act.

every 2-4 years founders relearn this lesson the hard way

@jayhinz 😭

@Sir_reads_a_lot @redditrepostss Yeah they're identical twins. Under normal circumstances they're going to have the same voice unless one twin trains it to be different or has a throat injury or something like that. have you never met an identical twin pair before?

@redditrepostss I suppose they had identical voices as well?

Why would you use Sol Ultra over Sol Max, given it’s a marginal gain in the coding index for what looks to be about a 35%ish increase in API cost? It literally looks like it’s a 1-point difference in the index between the two. Honest, for that matter, I would say Terra Ultra looks just as good for significantly less money. Looks like Sol Ultra is all of two points better on the coding index than Terra Ultra, but it’s 150% more expensive. I think that Luna Ultra seems to be the best performance-to-value. Looks like its score is 75ish, but at an API cost of $500, which means that Sol performs only 6% better than Luna Ultra but for 5x the price!

There are only 3 models: - Luna/High - Terra/Ultra - Sol/High

@pgerrits @rasbt Because software development isn’t the only use case for the models and this index metric is specifically measuring agentic coding.

@rasbt Sorry, but I find this so confusing. If the Luna High is better than a Sol Medium, why does the Sol medium exists? I'm honestly so confused with all these models and levels.

For agentic coding, one can say: - Unless you need Terra Ultra perf, it's always better to use a Luna model with higher effort setting (same or better performance but cheaper). - Forget everything below Sol High, use Luna with higher effort settings here - Forget Sol Extra High, use Terra Ultra here - The extra cost of Sol Ultra is probably not worth it over Max

This article explains exactly why it seems that the chat UX was entirely ignored; it’s because nearly everyone at OpenAI had already moved entirely to using Codex. Classic case of assuming all your users live in the same bubble you do. openai.com/index/how-agen… “Codex became the primary AI tool for every department at OpenAI. Engineering moved first, but Legal, Finance, and Recruiting crossed into Codex being their primary AI tool around April 2026. For the average OpenAI worker, Codex usage now accounts for more than 85% of output tokens. Since Codex users tend to use more tokens than non-users, its share of overall tokens is even higher: Codex accounts for 99.8% of weekly output tokens generated within OpenAI.”

@ahudovich It honestly seems like no one at the company used Chat because it’s so full of errors, but how can that be the case? Maybe because you can use Codex / Work to ask general questions too?

Chat experience in the new ChatGPT app is a total mess - No Projects (idk where they are now) - New chat opens in a side popup - If you want it fullscreen, it opens a new clumsy window - Text zoom doesn’t work - Global shortcut opens a task, not a new chat - Chat history is no longer in the sidebar It feels like they just duct-taped Chat onto Codex

@accelerate27 @NousResearch

@NousResearch Any pricing detail? Can't check to the "instance" unless I topup and subscribe to one of the plan

Hermes Agent is now in the Cloud! Setup couldn't be simpler: pick a model and a server size. Two clicks and 60 seconds later, your agent is live. Running a team? Spin up agents for everyone at your org with granular access controls and unified billing, all from Nous Portal.

@NousResearch hey @Teknium , is there a way to access the sandbox terminal backend? I feel like there's been some things using the "hermes" command that I could only accomplish in the terminal? I may be mistaken though...

@NousResearch AWESOOOOME

Fable and Mythos are the same model; Fable is Mythos but with more guardrails. "For a small group of cyberdefenders and infrastructure providers, we’re also launching Claude Mythos 5. It’s the same underlying model as Fable 5, but with the safeguards lifted in some areas." anthropic.com/news/claude-fa…

@theo I knew it will not at the level of Fable 5. The irony is Dario has two models ahead: Mythos 5>Fable5> Gpt-5.6. Dario is in beast mode.

Apparently I'm allowed to talk about GPT-5.6 now? It's a damn good model. Not quite as "smart" as Fable, but it is incredibly capable. Fixed all the problems I had with GPT-5.5. It is incredibly determined. Will run for a day without even using a /goal. It understands subagents incredibly well and is great at orchestrating. It's super pleasant in use cases like OpenClaw and Hermes Agent. It knows iOS dev incredibly well. It has rough edges too, but FAR fewer than 5.5 did. For many things, gpt-5.6-sol will become my obvious default. I will share a lot more soon 🫡

@AcunLK @OpenAI Google does tend to release a SOTA model that beats the others on benchmarks, but unfortunately for them, it only seems to last a couple of weeks before OAI or Anthropic releases a new one that leaves Google in the dust for months. Gemini Pro 2.5 and 3.0 were SOTA when released.

GPT-5.6 Sol, along with Terra and Luna, will launch publicly this Thursday. We’re expanding preview access globally now.