Man

[Man @MsShannon112]

@Lentils80 @ashen_one Claude models have more creativity and taste. It's been like this for a while now.

[Man @MsShannon112]

@jaiyagill His untrustworthiness helped build 2 unicorns. OAI and anthropic. Anthropic originated because Altman kept lying to people around him.

[Man @MsShannon112]

@iamtrask @ivanfioravanti Yep. And OpenAI was lobbying for no liability. Anthropic opposed them. https://t.co/PLDCWtXrNh

[Man @MsShannon112]

@yishan Did you read the article? The went back and looked over the past safety evals because of the HF incident. Timing of the this release makes perfect sense. The uncharitability people have for some Anthropic is crazy.

[Man @MsShannon112]

@SnehaRevanur This is the bare minimum

[Man @MsShannon112]

@neural_avb Apparently you have to prompt it differently.

[Man @MsShannon112]

@scaling01 Agree with altman led OpenAI who's sfatey lead resigned 1 week before the hacking incident and who has a poor history with saftey reaearchers who have ressigned and went over to anthropic. Dario and anthropic have been consistent from the start. They are more honest.

[Man @MsShannon112]

@JayaGup10 You are thinking about this with a poor mental of Anthropic. If this was Altman led OpenAI, sure he's a liar and no one trusts him. But Dario and Anthropic have been honest and deserve way more charity. I believe them.

[Man @MsShannon112]

@Lentils80 He might be trolling

[Man @MsShannon112]

@iamtrask It's implied.

[Man @MsShannon112]

@corsaren It's the smart thing to do. The play book by the bad actors like altman sacks a16z is to latch onto a small issue and magnify 100x. While when openai does anythign wrong it's brushed off. So no it's the correct approach to be as careful possible.

[Man @MsShannon112]

@kipperrii A way to know if someone is an OpenAI emplyee is to ask them if Altman is a liar. They will say no.

[Man @MsShannon112]

@ankitaich30 Good work. Do have a similar article for fable 5 and 5.6?

[Man @MsShannon112]

@jmbollenbacher listen to this https://t.co/FfQzdOsmPP

[Man @MsShannon112]

@chamath OpenAI didn't sign the AI alliance with Nvidia.

[Man @MsShannon112]

@gojomo It was just another altman lie lol. Dude hates safety people. The saftey lead resigned a week before the Hacking incident.

[Man @MsShannon112]

@SnorkelAI Do you have a performance and error analysis for fable and 5.6 like you did with opus 5?https://t.co/Wywh07VKDV

[Man @MsShannon112]

@ZypherHQ >However, even the simplest tasks become unnecessarily complex because it introduces changes that were never requested or even mentioned.

5.6 does this way more and write like 3 times more code which is un-PR-able. Opus 5 is actually really good.

[Man @MsShannon112]

@PjoyAI @AndrewCurran_ There is a massive anti-anthropic misinfo campaign by altman and his loyalists.