Josh

[Josh @Jqosh_X]

@MrZ2128 @MistralAI @grok I only use X for AI news and similar stuff, so I mostly use TikTok. What kind of argument is this? Just because you post and I don't? Bots can post too. Doesn't mean anything.

[Josh @Jqosh_X]

@MrZ2128 @MistralAI @grok You just know you lost the argument and are playing stupid now. or you are stupid, idk.

[Josh @Jqosh_X]

@MrZ2128 @MistralAI @grok I don't even use Mistral - I only have Claude Pro, GPT Pro, and SuperGrok. My point was never that Mistral Large 4 Preview is better than GLM 5.3, but in some cases it is! And it's a huge improvement for EU AI.

[Josh @Jqosh_X]

@MrZ2128 @MistralAI @grok IF YOU DON'T TRUST MY LINKS, GOOGLE IT YOURSELF—I SHOWED YOU HOW.

[Josh @Jqosh_X]

@MrZ2128 @MistralAI @grok I won't continue this argument. Your points make no sense. You know nothing and a 3-year-old is more capable. Pathetic. I don't post because I don't want to. Where are the arguments? None. You think there is only one benchmark and you don't know how to use Google.

[Josh @Jqosh_X]

@MrZ2128 @MistralAI @grok Google -> „Mistral AI“ -> Open the official website -> introducing Large 4 -> scroll down to Dense 200.

I can't believe you call yourself an „AI Influencer & AI Architect“ but you don't know shit and you talk like a bot.

[Josh @Jqosh_X]

@MrZ2128 @MistralAI @grok is my Link real and a good source?
https://t.co/zKrwDGZ3wh

[Josh @Jqosh_X]

@MrZ2128 @MistralAI 🫩🫩🫩
https://t.co/zKrwDGZ3wh

[Josh @Jqosh_X]

@MrZ2128 @MistralAI Not that hard to find

[Josh @Jqosh_X]

@MrZ2128 @MistralAI Are you really that slow? 😭 It doesn't say Astra there because that's the AutomationBench chart. Astra is in the VISION benchmark (Dense 200). Different chart, different benchmark. That's literally what "many benchmarks" means.

[Josh @Jqosh_X]

@MrZ2128 @MistralAI Exactly. Astra is closed and one of the best models out there. And an open model beats it on vision. That makes it MORE impressive, not less. Thanks for making my point.

[Josh @Jqosh_X]

@MrZ2128 @MistralAI Okay, slowly: this is ONE chart. In this one, GLM is a tiny bit better. 62 > 59, yes, good job. But there are OTHER charts too. In vision, cyber and price, Large 4 wins. And Mistral went from 6 to 60 in one jump. That's called progress. Understood now, buddy?

[Josh @Jqosh_X]

@MrZ2128 @MistralAI Still on 62 > 59 a day later? Nobody disputed that. 15 replies, zero arguments, can't even spell "you're," and you call me the idiot? Embarrassing. I feel like you didn't even read one post of mine. This has to be ragebait.

[Josh @Jqosh_X]

@Arnobrzh @SidharthVijay_ Not terrible - 100% a huge improvement for the EU. Beats K3 (Max) on Auto Bench & Term Bench, much cheaper and faster per task, and beats costlier K3 (low) almost everywhere. So there are many use cases where it’s a lot better than K3. And it’s still a Preview.

[Josh @Jqosh_X]

@MrZ2128 @MistralAI Bro, you've replied like 10 times and made exactly one point, which is 62 > 59, something nobody ever disputed. Can't spell "you're", can't read a tweet, and keeps spamming the same line every minute. Genuinely shut the fuck up 🙏

[Josh @Jqosh_X]

@MrZ2128 @MistralAI Never said 59 > 62. Large 4 is open-weight too, and Astra is only the vision comparison because GLM 5.3 can't do vision at all. Also no idea why you're suddenly talking about the AI Act.

[Josh @Jqosh_X]

@MrZ2128 @MistralAI Why are you so mad? I said "a lot of", not this one. Large 4 beats GPT-6 Astra on vision, tops the AA Cyber Index and is way cheaper per task than GLM 5.3, so it's huge progress for EU AI and simply the better pick for a lot of use cases and not a joke.

[Josh @Jqosh_X]

@MrZ2128 @MistralAI Not everywhere. Large 4 Preview has Astra-level vision and comes close to Opus in many benchmarks, crushing GLM 5.3.

[Josh @Jqosh_X]

@MistralAI OK, I'll switch to le chonk🥹