Gerard Smit

@github This is insane. "stand by"? As in please subscripe to pro and pro+, come back to us after we bait and switch changing our subscription model and we'll do it again? No thanks. I'll keep my Claude and OpenAI subscription.

Weekends are for building. Copilot Max users, check your account for an extra $200 in credits to power your next build in the GitHub Copilot app. Stand by for more offers for Pro and Pro+ users.

@Cyberw0t @ns123abc No it's electricity production: en.wikipedia.org/wiki/List_of_c… But yes. It's a lot of coal. Back in 2021 there was a flood in a coal mine which dipped energy production, causing Bitcoin hash rate to go down: decrypt.co/68591/flooded-… But we have to make cleaner energy 🤷‍♂️

@socialgerard @ns123abc Is this a chart that shows air pollution cause that’s a lot of coal.

🚨BREAKING: TRUMP ON ANTHROPIC REPORTER: Do you view Anthropic and to a degree its CEO, Dario Amodei, as a threat to national security? TRUMP: "Well, not now, but a week ago, maybe. I was with him yesterday, he made a speech... seemed like a nice guy, smart guy. But he responded to us very quickly, because you know it's a tremendous liability… he responded very responsibly, I thought." REPORTER: Would you consider using the Defense Production Act to possibly regulate or control AI? TRUMP: "I would, but, I'm not sure I have to do that. I think so far it's been very responsible. Actually, it was a competitor, and a part-owner, that turned Anthropic in…"

@JaidCodes Their cache is broken. I used them directly, and received this e-mail. They wanted me to pay up for the mistake they made. x.com/socialgerard/s…

@arrakis_ai Average time: 43 minutes and twice as expensive than GPT-5.5 is kind of insane...

Claude Opus 4.8 has landed on DeepSWE Bench, posting a 58% Pass@1 and taking #2 overall behind GPT-5.5. It continues a broader trend: slightly behind on raw score, but among the most reliable and efficient coding models across recent benchmarks.

When you're trying to mention a GitHub Action workflow to Claude: don't forget to press [Alt] + [W] (Alt = meta in Windows) to NOT use the new workflow feature. Or say "w-orkflow"...

@Cointelegraph Keep in mind that Mythos Preview is priced at $25/$125 per million input/output tokens. That's 5x more expensive than Opus. So it's probably not going to be cheap.

🚨 UPDATE: Anthropic says it expects to bring Mythos-class models to all customers in the coming weeks.

@llmdevguy Anthropic released a new model, and have a better model coming up the next weeks (Mythos), then I'm expecting something better than this when I try it out.. 😅 x.com/socialgerard/s…

Did Opus 4.8 beat GPT-5.5? Please tell me because I don't use anything from Anthropic.

Getting the same in Claude CLI (new session):

Trying to continue doesn't work either. The session is cooked.

First prompt in Opus 4.8 in Claude Desktop. Ohno:

@saeed_vz Did you see the set-up fees of new dedicated servers? If you buy a dedicated server now, it's a commitment...

Not again hetzner :(

@wstaelens @James_M_South Foo is [var item, ..] ? item : default; 😉

@James_M_South love this replacement for Foo.FirstOrDefault(): Foo is { Count: > 0 } list ? list[0] : default;

if (a is { } b) { } This is the worst C# pattern. I absolutely hate it with all my blackened heart. It's unreadable crap.

@Leapofcode Yeah, a couple year ago. But it was in the browser, not a seperate app.

Google has launched its Google Search Desktop app. Have you tried it?

@BrianRoemmele They can still use Claude models in Copilot. The difference is that Microsoft now can use the conversations to train their own AI model. They already have finetuned models like Raptor mini and Goldeneye. So maybe we will see more in the future.

MICROSOFT CANCELED CLAUDE CODE! IT COST TOO MUCH. Major tech companies are confronting the steep reality of AI inference costs as the era of heavy subsidies appears to be ending. Microsoft is canceling most internal licenses for Anthropic’s popular Claude Code tool by June 30, 2026 less than six months after rolling it out broadly to engineers primarily due to escalating token-based expenses, while shifting teams toward its own GitHub Copilot CLI. This mirrors broader pressures: Uber’s CTO revealed the company had already exhausted its entire 2026 AI budget in just four months thanks to heavy Claude usage among thousands of engineers (with individual monthly costs often hitting $500–$2,000), and GitHub is transitioning Copilot to usage-based billing with higher per-token rates starting June 1st. The reality is good enough AI will expand and constantly get better removing the oxygen of the most expensive.

@DeepInfra Also not saying sorry but only "thanks for using deepinfra" - now I see why people are using @OpenRouter ...

Huh? So there was a mistake in the cached token count billing, I still have balance left in my account and now @DeepInfra wants me to pay up for their mistake? I was playing around with OpenClaw. If I knew it would've cost me 50 eur more, I wouldn't have topped up.

@VincentBuilds @bridgemindai Don't bother replying. They are probs using an AI to respond to farm interactions; they even ran it twice accidently 🤷‍♂️

@bridgemindai But they replied to you directly and offered help tho

Just cancelled my Google AI Ultra subscription. Google scammed me and I'm done. I bought AI Ultra for $250/month the day before Google I/O specifically to test the new models. Then Google dropped the price right after I paid full price. No refund. No credit. Nothing. Then the model they released? Gemini 3.5 Flash. #14 on BridgeBench. #22 in reasoning. Produces slop on real tasks. I will never use this in a production codebase. I paid a premium price for a model I'll never use and Google won't make it right. This is the second time I've cancelled Google AI Ultra. It's the last.

@ZooL_Smith I'm not sure if it works in Godot, but there is a pauseless GC called Satori for .NET. It's an experiment by a Microsoft dev: #discussioncomment-14263828" target="_blank" rel="nofollow noopener">github.com/dotnet/runtime…

I'm done with .NET on Godot, I'll be migrating to either gdscript or cpp. I've been chasing 2 frames hitch issues for a week, most of the API returning objects are causing the heap to climb a lot and the GC cleaning em creates that hitch. I'm out of options. Do not use C#.

@TheCyberSecExp @danveloper I guess this is the "you have memories" turned off image. I got the same-ish image.

@danveloper Kind of a theme

Try this prompt in ChatGPT: "Based on everything you know about me generate an image I can use as my wallpaper for my laptop that represents who I uniquely am. Do not overcomplicate it visually. Abstract representation is fine. No words."