Ryan James (@beezlebuddy) - Twitter Profili | Zamantika Mersobahis Locabet

Ryan James@beezlebuddy·20h

@zcbel31 @NeubyNdfb @MathiasTheus @bucketshopcap call me out in a few months/September if I'm directionally massively wrong x.com/beezlebuddy/st…

Ryan James@beezlebuddy

@dewmboom @MicronTech hi there I've been following MU since last year my prophecy is thus: MU will go up leading to the earnings release earnings will hit it will drop 8-9% and take the next week to find it's footing and then end the second week after earnings (first part of July) up 15%

English

0

32

ZC@zcbel31·1d

@NeubyNdfb @MathiasTheus @bucketshopcap Not a fat, but it will take longer to implement and advance than is currently priced in. Markets and equities don’t go up in a straight line. In fact, watch what happens to $MU price tomorrow and next after OPEX

English

1

0

1

148

Bucket Shop Capital@bucketshopcap·1d

$MU is literally free money IMO and hilarious that anyone would own literally any stock over having every available free $ in $MU.

Evergreen@evrgn11112231

Still true, literally free money IMO and hilarious that anyone would own literally any stock over having every available free $ in $META.

English

14

4

276

71.5K

Ryan James@beezlebuddy·1d

@thsottiaux @moritzkremb openai has awesome models but claude still has the better terminal harness. I wish you all could use claude code so you can see what it's like. Its beautiful.

English

0

142

Tibo@thsottiaux·1d

@moritzkremb Let’s see. Maybe that’s also our strategy 🤔

English

35

0

625

16.8K

Moritz Kremb@moritzkremb·2d

Feels like everyone in the Twitter AI bubble is switching to Codex lately But I think Anthropic will release something new soon, like a new model, way better app, or crazy new features and then everyone will just flock back

English

116

2

321

36.2K

Ryan James@beezlebuddy·1d

@Erik_Hamre @agmethod75 @MarkovMagnifico danish do know how to build bike infra and they bike in all weather like champs

English

2

0

1

79

Erik Haalo@Erik_Hamre·1d

@agmethod75 @MarkovMagnifico 10 minutes bikeride from where I used to live, which was 10 minutes bikeride from Copenhagen center.

English

5

0

2

856

Markov@MarkovMagnifico·1d

holy shit there’s no way the UK is actually this poor

English

456

118

10.9K

2M

Ryan James@beezlebuddy·1d

@thsottiaux please don't take away our headless codex prompts :(

English

0

1

257

Tibo@thsottiaux·1d

They meant companies... 2000 companies. We're going to need to stand up a lot of compute 😅

OpenAI Developers@OpenAIDevs

2000 developers reached out in 3 hours. Let's build things.

English

84

39

1.7K

156.6K

Ryan James@beezlebuddy·1d

@dewmboom @MicronTech none of this is financial advice/guidance its just me having fun call me out if I'm wrong! 😇

English

0

9

Ryan James@beezlebuddy·1d

@dewmboom @MicronTech on June 23 it will be $931 on June 26 it will be $850 on July 1 it will start bouncing around $1000 but struggle to mount it EOD by July 15th it will be at $1050 slowly creeping up and sometimes down until the FY2026 release in September where it will end at $1200 mid September

English

1

0

13

DewmBoom@dewmboom·5d

$MU POLL: Vote Below 👇🏼👇🏼 @MicronTech is expected to announce Q3 earnings on June 24. Where do you think the stock price will be at by end of June? >: Larger Than <: Smaller Than

English

3

1

15

11.5K

Ryan James@beezlebuddy·2d

@htihle damn I may have to rethink my go-to opus model glad to see the goat gpt 5.5 xhigh is where I would expect

English

0

203

Håvard Ihle@htihle·2d

Claude Opus 4.7 (max) scores 75.4% on WeirdML, not exceeding 4.7 (high) or (no thinking). The error bars are large, but it appears that reasoning tokens do not help much for Opus 4.7 on these tasks (unlike Opus 4.6). It uses a lot of tokens, and is more than 4 times as expensive as 4.7 (high).

Håvard Ihle@htihle

GPT 5.5 (xhigh) scores 84.9% on WeirdML taking the lead over 5.5 (high) 83.9%. Even (xhigh) is not using more than about 15k output tokens. Thanks to @METR_Evals for the increased support that allowed for this run. Opus 4.7 (max) soon, and more things in the pipeline.

English

6

5

77

7.2K

Ryan James@beezlebuddy·3d

@htihle @METR_Evals also I feel like I'm in the minority that sees improvement in opus 4.7 max over 4.6 at any reasoning level, want to know if that is just some kind of bias on my part

English

0

1

34

Ryan James@beezlebuddy·3d

@htihle @METR_Evals awesome, cant wait to see these results vs opus 4.7 max, my two go to model + reasoning level combos

English

1

0

1

225

Håvard Ihle@htihle·3d

GPT 5.5 (xhigh) scores 84.9% on WeirdML taking the lead over 5.5 (high) 83.9%. Even (xhigh) is not using more than about 15k output tokens. Thanks to @METR_Evals for the increased support that allowed for this run. Opus 4.7 (max) soon, and more things in the pipeline.

Håvard Ihle@htihle

WeirdML v2 is now out! The update includes a bunch of new tasks (now 19 tasks total, up from 6), and results from all the latest models. We now also track api costs and other metadata which give more insight into the different models. The new results are shown in these two figures. The first one shows an overview of the overall results as well as the results on individual tasks, in addition to various metadata. The second figure shows cost vs performance and shows a clear scaling with better results for higher costs. We also have a very varied pareto frontier with 11 models from 6 different companies having the best accuracy for a given cost for at least some of the cost range. Grok 3, Claude Opus 4 and GPT 4.5 are the ones that underperform for their costs, while Gemini pro and o3 pro have the best results at the highest costs. Qwen3 30B3A, grok 3 mini and deepseek R1 also each represent a good chunk of the pareto frontier.

English

6

12

186

27.9K

Ryan James@beezlebuddy·3d

@dewmboom micron is entering the hype-o-sphere which means big jumps up but also big jump downs, it’s going to get weird for the rest of 2026 i think

English

0

537

DewmBoom@dewmboom·3d

$MU needs some rest before the leg up over $1000. Let it rest a little!

English

8

4

185

7.5K

Ryan James@beezlebuddy·7 May

@ecommerceshares my point being, europeans (in my experience) tend to be more techno skeptical and ai skeptical especially and I think that this extrapolated out to the population means less ai being pursued

English

0

1

7

Ryan James@beezlebuddy·7 May

@ecommerceshares anecdotal take, based on my experience living in Europe and working for a Danish company Danish people -- very excited about it (not great sample as all these people are coworkers I work with) everyone else I talk to, suspects it's going to poison the water and destroy all life

English

1

0

2

80

Wasteland Capital@ecommerceshares·7 May

Why has Europe decided to completely stay out of artificial intelligence race? Serious question. They’re not even trying. I really don’t get it.

English

2.7K

200

7K

1.1M

Ryan James retweetledi

Zach Brock@z·7 May

congrats to anthropic for defeating grok in the market and feasting upon the compute of their fallen enemy

English

77

192

4.1K

168K

Ryan James@beezlebuddy·5 May

@diegohaz I have Opus 4.7 max build a plan. I have GPT 5.5 xhigh look at it and rewrite it. Then I ask Opus 4.7 what it think's of thew plan. Oh yes it's better. Or sometimes have Gpt 5.5 just code Opus 4.7's plan. Always the same: "Yeah this is a lot better than I planned."

English

1

0

61

Haz@diegohaz·5 May

I gave the same task to GPT-5.5 xhigh and Opus 4.7 max. – GPT took ~30 minutes. – Opus took ~2 hours. Then I asked them to review each other's work and give an honest verdict on which was better. – GPT said its own code was better. – Opus said there was no obvious winner. Then I asked them to learn from each other and apply the best parts they had learned to their own code. In the end, I asked them to review each other again and give an honest verdict after they had both improved. – GPT kept saying its own code was better. – Opus said GPT's code was better. What a journey.

English

187

77

3.1K

346K

Ryan James retweetledi

Paul Graham@paulg·29 Nis

Trump's bullying only works (and only temporarily at that) because he's taking advantage of the high-trust customs and relationships established by more principled predecessors.

English

149

212

3.9K

198.4K

Ryan James@beezlebuddy·27 Nis

@hispanicnomad @joehas @es_INE estoy aqui en espana soy injiero la verdad :) gracias

Español

1

0

1

502

Hispanic Nomad | Remote Work, Travel, Growth@hispanicnomad·27 Nis

@joehas @es_INE Yeah we are exporting qualified professionals and importing “doctors and engineers” from Somalia I’m sure that’s what Spanish people want

English

7

3

233

7K

Hispanic Nomad | Remote Work, Travel, Growth@hispanicnomad·27 Nis

Spain has better weather than every Nordic country, better food than France, better nightlife than Germany And somehow managed to make it so painful to live there, that people are leaving en masse

English

229

242

4.9K

181.8K

Ryan James@beezlebuddy·27 Nis

@htihle any chance we will see where high and xhigh gpt 5.5 fit?

English

1

0

6

801

Håvard Ihle@htihle·27 Nis

GPT 5.5 (no thinking) scores 67.1% on WeirdML, well ahead of GPT 5.4 (no thinking) at 57.4%, but well behind Opus 4.7 (no thinking) at 76.4%. It's at the frontier for accuracy/tokens, as it uses less tokens than Opus.

Håvard Ihle@htihle

WeirdML v2 is now out! The update includes a bunch of new tasks (now 19 tasks total, up from 6), and results from all the latest models. We now also track api costs and other metadata which give more insight into the different models. The new results are shown in these two figures. The first one shows an overview of the overall results as well as the results on individual tasks, in addition to various metadata. The second figure shows cost vs performance and shows a clear scaling with better results for higher costs. We also have a very varied pareto frontier with 11 models from 6 different companies having the best accuracy for a given cost for at least some of the cost range. Grok 3, Claude Opus 4 and GPT 4.5 are the ones that underperform for their costs, while Gemini pro and o3 pro have the best results at the highest costs. Qwen3 30B3A, grok 3 mini and deepseek R1 also each represent a good chunk of the pareto frontier.

English

13

155

66.4K

Ryan James@beezlebuddy·26 Nis

@scaling01 > for all their past reckless actions like what? if it were up to Anthropic we would still have zero public models because autistic spectrum Dario would still be telling everyone there is no safe model until he alone deems it so idk, I guess I just generally disagree

English

0

5

169

Lisan al Gaib@scaling01·26 Nis

Nothing changed for me. To be very clear: I still want Anthropic to win. They are taking the safer and more principled approach and I trust them a lot more. But I'm not going to sit here and pretend GPT-5.5 or any other OpenAI model sucks because of that. Since GPT-5.2-xhigh I have been saying that OpenAI has smarter models, and this trend has continued. My only complain was the reasoning-efficiency, but they fixed that. I have also made predictions about OpenAI pulling ahead way before all of this happened. Saying their models are better or enjoying Sam's recent drunk comments or post doesn't mean I agree or forgive them for all their past reckless actions. But I do appreciate Sam coming forward. It shows good will. Of course I'm not blind to why Sam might be doing all of this. The recent personal attacks on Sam's home, the upcoming Elon vs OpenAI trial and growing anti-AI sentiment are very good reasons to up your PR game and reflect on your actions. I would do the same. So I'm not discounting the possibility that he's doing this to save his own skin, but I also don't want to discount the possibility of this being a genuine attempt to fix things.

English

54

12

519

46.6K

Ryan James

Keşfet