@Govindtwtt I feel they aren't talking it seriously and their focus is probably enterprises how use Gemini in google products that's where money is. Also they'll be subsidizing a lot if they create some high end frontier models so they just do cheap ones and make it usable in their products
@usman_akinyemi Won't give you full Linux experience but could teach you somethings about Linux. I still use it even though I dual boot cause ion wanna dual boot everytime or run a heavy vm just to use docker or other Linux tools
Restful aren't going anywhere. If we allow AI agent to decide then we move from a deterministic approach to a non deterministic one. And you'll burn tokens on something totally unnecessary. I know AI agents are the future but this isn't where they belong. Or else CHAOS 😂
RESTful APIs may be dead soon. Instead, web services may expose a single POST entry point for a prompt. Internally, an AI agent may decide how to interpret it and what to do with the data and the database.
@afifsohaili Yes, it actually does. IDM splits files into multiple parallel connections, bypasses server-side throttling, and resumes broken downloads. Not just animations lol
@wikiwayne@Hartdrawss Background jobs (Nextjs api route aren't designed for long running tasks due to timeouts and all) , connection pool limits, or team size >5.
This is built into the harness, not the model. The fake tool injection (and reasoning summarization) is gated behind internal‑only flags likely to prevent Anthropic employees and beta testers from using Claude outputs to train rival models, whether intentionally or through leaks
I cannot stop thinking about Anthropic today for some reason
1. They claim that they are a company that prioritizes safety first and that they are creating a model responsibly
2. we learned from the code leak that anthropic employs deceptive techniques by calling fake tools to throw off distillers...
Is this lying pattern built into claude or just the harness running claude? What else are they lying about? I am a bit more concerned now.
@ThePrimeagen The real concern isn’t deception of customers, but lack of transparency: Anthropic hasn’t disclosed these anti‑distillation measures. If they’ll silently alter outputs for one purpose, what else might they filter or change without telling us?
@ThePrimeagen This isn’t the model lying – it’s the harness (API/CLI wrapper) adding fake tool calls after generation, only for certain internal flags. Normal users don’t see it.
@akinkunmi A separate subdomain gives you a different origin, which can improve isolation.
A /admin path is not the same boundary. But subdomains are only safer if cookies and auth are scoped correctly; otherwise you can still create shared-risk problems.
@Hesamation This probably didn't get adopted because json is known for its simplicity and also most tools already have battle tested json library and doesn't need external schemas to understand the data.
But still his was a great project
@theo Firstly tauri don't use so much ram it does that if the front-end Isn't optimized (most time if react re-renders)
Tauri is crazy fast and small app size and works with webview unlike electron which boots up chromium to work. So tauri just has a smaller ecosystem (trade-off)