@ZFingerhut14747 @iruletheworldmo He's a scammer, capitalizing on your engagement. What do you think?
English
Ridhvik Gopal (Vik)
36 posts
@iruletheworldmo you lied about having insider knowledge on spud >:(
English
@haider1 I'm not sure benchmarks are a great indicator of model size. They have a new tokenizer. Possible that they have a new, slightly larger base model after learning something from the Mythos pretraining. Maybe Opus 5 will be a refined version of 4.7 with new post training techniques
English
from what i understand, anthropic won't release 'mythos' anytime soon
opus 4.7 is probably a smaller model, which would explain the lower benchmarks in some areas
that is a good thing, because anthropic can't serve opus 4.6 with the amount of compute they have right now
English
@abhishekray Thoroughly enjoyed reading this. Thanks!
Is it possible they have a new base model too?
Seems like with Mythos they did a big pre-training run. Maybe they learned something and are slightly scaling up their Opus model as well?
English
Ridhvik Gopal (Vik) retweetledi
@bcherny @eliebakouch “stacks random distractors to "trick" the model, unlike real long-context use (e.g., code/docs).”
Isn’t it a good thing that the model cannot be distracted and isn’t it a good indication of quality- actually understanding the long context enough to not be distracted/confused?
English
👋 We kept MRCR in the system card for scientific honesty, but we've actually been phasing it out slowly.
Two reasons: (1) it's built around stacking distractors to trick the model, which isn't how people actually use long context, and (2) we care more about applied long-context capability than needle-retrieval. Graphwalks is a better signal for applied reasoning over long context, and internally we've seen this model do really well on long-context code.
MRCR wasn't included in the Mythos Preview system card for these reasons, but Graphwalks was - that will be the case for future models too.
English
wait why is there an INSANE gap on long context benchmarks between opus 4.6 and 4.7??? this is crazy
Claude@claudeai
Introducing Claude Opus 4.7, our most capable Opus model yet. It handles long-running tasks with more rigor, follows instructions more precisely, and verifies its own outputs before reporting back. You can hand off your hardest work with less supervision.
English
Ridhvik Gopal (Vik) retweetledi
@LexnLin Again? Has he ever been right?
All these guys do is tweet “it’s coming tomorrow” every single day. A broken clock
English
i bet that strawberry man is wrong again
we wont get all three today
(superapp, spud, image v2)
lets just hope for one
🍓🍓🍓@iruletheworldmo
happy release day one and all. this will go down as the most monumentous day in human history. (it’s a word, go away) it’s one small step for man, and one mighty giant leap for ai. happy release day to all that celebrate 🥹 what upcoming release excited you most? -the SUPER APP -the SPUD -the IMAGE-GEN for me. it’s all about the glorious combination.
English
@Postulix96 @synthwavedd You do know most of these “influencers” are just capitalizing on your engagement right?
They keep tweeting “it’s coming tomorrow” vaguely and when something comes, they’ll say “see I told you so!”
English
@christiangori96 *clears throat*
“They say AI killed LinkedIn.
I’m pleased to announce that I disagree.
It exposed the lazy.
While you complain about bots, I’m leveraging LLMs to 10x my authentic empathy and scale my vulnerability and learn lessons on B2B marketing”
/s
English
@yacineMTB And then watch all their servers simultaneously melt and throw a tantrum.
English
@Vera28765582815 @TheRealAdamG They’ve designed it to be steerable and great at instruction following. The downside is that for this model, that means it’s like a smart toddler that needs to be told EVERYTHING and how to do it, and becomes bad at inferring intent from context.
English
@TheRealAdamG 5.4 isn’t great at understanding what the question is or what would be the most helpful. But once it start working on the right task can it be good!
English
@thsottiaux Is there any difference in usage for accessing the 5.4 pro model in the $100 vs $200 tier in ChatGPT? The terms are silent on this, but a lot of people would like to know.
English
@thsottiaux Can you clarify what it means to the Pro model in ChatGPT? What are the limits there?
English
Hi! Getting a lot of questions on Pro plan rate limits, so I wanted to clarify a few things:
- Your Pro $100 plan includes at least 10x Plus usage, till May 31 with the 2X usage boost.
- Your Pro $200 plan includes at least 20x Plus usage, till May 31 with the 2X usage boost. This is also the SAME usage this plan had since the 2x promo in February; we previously didn't document this explicitly.
Now to address the confusion. Our pricing page says "5x or 20x usage", and many of you very understandably took it to mean "10x and 40x" given then 2X usage boost. That's our fault. We mixed up two things in a confusing way on that page - (1) sharing that the Pro $100 launched with a 2x boost, making it 10x Plus till May 31 and (2) sharing that Pro $200 retained its 2x boost and sharing for the first time that this equates to 20x Plus.
We're going to update our pricing pages to make this clearer. Sorry it wasn't as clear as it should have in the first place.🫥
am.will@LLMJunky
There seems to be some confusion on what ChatGPT plans give you, and what bonuses are active right now. I got you. ChatGPT Plus $20 = 1x NO ACTIVE BONUS ChatGPT Pro Lite $100 = 5x with a 2x Bonus (10x) ChatGPT Pro $200 = 20x with a 2x Bonus (40x) The bonus is ONLY for Pro plans. It includes BOTH $100 & $200 Tiers They both end May 31st. THANK YOU FOR YOUR ATTENTION TO THIS MATTER. SIGNED AMW
English
@theo @khan_kilo The gap between top performers and average ones is widening into a visible class divide...just like everywhere else. I hope we build safety nets so no one gets left behind.
English
There's pretty much no reason to hire a mid tier developer now
English
Ridhvik Gopal (Vik) retweetledi
NVIDIA CEO watching everyone build their own chips
Google. Meta. Amazon. xAI and now ANTHROPIC 😭
Polymarket@Polymarket
JUST IN: Anthropic reportedly considering designing its own AI chips to reduce dependence on Nvidia.
English
@Angaisb_ Yeah, but “usage allowance” could also mean access to the Pro model on ChatGPT.
There is no clarity in usage difference between the two tiers there
English
There is also nothing about it in the FAQ, just this: "Both Pro plans include the same core capabilities. The main difference is usage allowance: Pro $100 unlocks 5x higher usage than Plus (and 10x Codex usage vs. Plus for a limited time), while Pro $200 unlocks 20x usage than Plus."
English
Many people seem to be confused but it's easy:
- $100 Pro Plan is the same as the $200 Pro Plan in ChatGPT, no differences there
- $100 Pro Plan gives you 5x usage in Codex
- $200 Pro Plan gives you 20x usage in Codex
- Both Pro plans keep the 2x usage promo until May 31st
Angel 🌼@Angaisb_
ChatGPT Pro $100 plan is out! At least we got something today
English
@OpenAI How is usage defined in the Pro model?
I really like that to make my money expense budgets financial models and fun research questions, but there is no guidance on the difference in usage between pro 100 and 200, for the GPT-Pro model.
English
We’re updating our ChatGPT Pro and Plus subscriptions to better support the growing use of Codex.
We’re introducing a new $100/month Pro tier. This new tier offers 5x more Codex usage than Plus and is best for longer, high-effort Codex sessions.
In ChatGPT, this new Pro tier still offers access to all Pro features, including the exclusive Pro model and unlimited access to Instant and Thinking models.
To celebrate the launch, we’re increasing Codex usage for a limited time through May 31st so that Pro $100 subscribers get up to 10x usage of ChatGPT Plus on Codex to build your most ambitious ideas.
English
@ShaneLegg @ishanhaq I read Tick-Tock-tick-tock-Bing couple of month ago and discussed with a colleague about how you pretty much nailed it!
Love it when leaders in AI are scientists and engineers first and 'management-consultants' only when absolutely necessary.
English
@ishanhaq See the end of this post from 2009
vetta.org/2009/12/tick-t…
English
