Sarjan Narwan

[Sarjan Narwan @sarjnarw]

@giffmana @trq212 I swear this is every PR situation with Anthropic. Cut rate limits, then after people start talking about it say it's to deal with load.
Just say what the problem is.

[Sarjan Narwan @sarjnarw]

@NoemiTitarenco @krishdotdev We want AI agents that can discover like we can, not which contain what we have discovered. Building in our discoveries only makes it harder to see how the discovering process can be done.

It's from the bitter lesson by Richard Sutton. The OG RL guys.

[Sarjan Narwan @sarjnarw]

@SenSanders Wealth taxes are one of the dumbest taxes out there. They fail because of price discovery and liquidity.
Capital gains / getting rid of the debt trick to avoid capital gains are what should be the focus.

[Sarjan Narwan @sarjnarw]

@CounterTyrannyX @EvanKirstel @DavidSacks Gemini is one of the worst models out there for coding, hallucinated a bunch and fails at tool calls often. Depends what your use case is.

[Sarjan Narwan @sarjnarw]

@PolybiusChamp @joshuarolson @RBPundit This is Loki’s paradox. Just because it’s difficult to draw a line doesn’t mean you can’t make a decision. I’d say it depends how much it promotes things like having kids especially with fertility rates in western countries.

[Sarjan Narwan @sarjnarw]

@therealdelairj @thsottiaux @maxedapps @AnthropicAI same 😅

[Sarjan Narwan @sarjnarw]

@theo @maria_rcks nahcrof have been close to 50% off on most open models.

https://t.co/QQSy4n9usp

[Sarjan Narwan @sarjnarw]

@theo tokens/s I don't think is fair to say until it's released and there's more providers and pricing will drop when other providers host it.

[Sarjan Narwan @sarjnarw]

@dhh Net zero is stupid, the argument should be renewable to reduce reliance on other countries that may be unreliable. (Russia / Straight of Hormuz)

[Sarjan Narwan @sarjnarw]

@Niwrax @thsottiaux It's fine, but I think some subtle visual indicator would be nice.

[Sarjan Narwan @sarjnarw]

@jimcramer You know what an "Open" model is right? They don't have your data, the risk is model poisoning.

[Sarjan Narwan @sarjnarw]

@MarcosHernanz Not yet.

[Sarjan Narwan @sarjnarw]

@gm_mertd They're great even if you don't use them as they force closed providers to bring prices down / to innovate more.

[Sarjan Narwan @sarjnarw]

@CodeByNZ Atleast it was a no comment, also when the weights become open these bias's are normally trained out.
https://t.co/K7BC9WbLHe

[Sarjan Narwan @sarjnarw]

@WatchChad Calm down

[Sarjan Narwan @sarjnarw]

@shiri_shh Margins / Revenue / In China (capital controls etc)

[Sarjan Narwan @sarjnarw]

Might take me a few years for us to pay this one off @aws

[Sarjan Narwan @sarjnarw]

@thsottiaux Maybe some visual indicator to indicate which is which? (if in all)

[Sarjan Narwan @sarjnarw]

@_12184 @0xAlaric That's cost from Moonshot, with open models you'd expect different pricing from other providers.

[Sarjan Narwan @sarjnarw]

@heyandras @Jean_build Probably wouldn't be good for caching.