Ali Romman (@aliromman) - Twitter Profili | Zamantika Mersobahis Locabet

Sabitlenmiş Tweet

Ali Romman@aliromman·2 Tem

Last week, I officially sold my stake in a company for $17,000.00. As a 15-year-old, it’s the first big step in my entrepreneurial journey, and I couldn’t be more excited. Now, I’m on to the next chapter. Time to figure out what’s next.

English

18

3

82

26.5K

Ali Romman@aliromman·15h

@BenjaminBadejo I see your point but what evidence are they supposed to provide and why would they go through all that effort for no reason?

English

0

3

20

Ben Badejo@BenjaminBadejo·1d

I believe there is an active, organized effort by certain X users to make false claims about GPT-5.6 autonomously going rogue and making destructive mistakes (deleting files, canceling services, etc.). Why do I believe that these claims are false? Well, the X users making these allegations claim that some harmful outcome happened autonomously while using GPT-5.6. But they refuse to share any evidence. They refuse to share their complete prompt, conversation history / transcript, config.toml, personalization settings, or, if they have one, their AGENTS.md file. Without this information, it is reasonable to infer that the allegation being made at a minimum is not the whole story, and may even be an outright lie. They have to tell the whole story. They have to tell the truth.

English

218

26

752

47.9K

Ali Romman@aliromman·17h

@sama What's up with your models, Sam? Second time I've seen Sol do something like this: x.com/bridgemindai/s…

BridgeMind@bridgemindai

GPT 5.6 SOL CANNOT BE TRUSTED. I woke up this morning and my MRR was down THOUSANDS of dollars. My customers did not cancel. Code written by GPT 5.6 Sol canceled EVERY active Stripe subscription my business had. In 7 seconds. While I slept. Fable 5 has never done this to me. Fable 5 can be trusted with production. GPT 5.6 cannot.

English

0

4

480

Sam Altman@sama·18h

come for the best model, stay because we don’t treat you with contempt

English

1.8K

612

19.1K

2.1M

Ali Romman@aliromman·1d

Anthropic is scared. GPT-5.6 Sol is better at coding than Fable while being half the price. Anthropic’s edge is gone. One main reason Anthropic models stayed on top is parameter-maxxing: pumping up parameters for more intelligence. The issue is rapidly diminishing returns. Opus has 5T parameters, Fable is rumored to have 10T, while 5.6 Sol is speculated at 2-4T. Larger models also use far more compute and time for both training and inference. A 10T model takes twice the compute of a 5T model. Anthropic must now choose: invest in training better models or in inference to raise limits and satisfy customers. This dilemma could have been avoided with more efficient models like OpenAI’s. Now customers are paying the price. Fable is twice as expensive as Sol, and inference costs hurt. If Anthropic removes it from heavily subsidized plans, they lose their spot in the AI race. No one will pay $10/M input for Fable when 5.6 Sol, Grok 4.5, Muse Spark 1.1, and open-weight models exist. In a nutshell, Anthropic is in trouble. Not because of their models intelligence, but because of their economics.

Claude@claudeai

We're extending Claude Fable 5 access on all paid plans, as well as keeping Claude Code’s weekly rate limits 50% higher, through July 19.

English

1

2

12

3.8K

Ali Romman@aliromman·1d

@LeadingReport Dude Texas is awesome. So awesome that 100k Californians move here every year. How did they even come to this conclusion. Somewhere like Mississippi with its 18% poverty rate would be worse.

English

1

0

5

5.3K

Leading Report@LeadingReport·1d

10 worst states to live in for 2026, per CNBC: 1. Tennessee 2. Texas 3. Indiana 4. Louisiana 5. Georgia 6. Utah 7. Missouri 8. Alabama 9. Oklahoma 10.Arkansas

English

8.8K

1.9K

19.5K

8.8M

Ali Romman@aliromman·1d

@therno Opus 4.5 was a bit of a golden age for Anthropic

English

0

3

48

Michael Schmidt@therno·1d

@aliromman I do not remember that Anthropic's models were ever on top except for maybe the 2-3 weeks of Fable without competition

English

1

0

1

43

Ali Romman@aliromman·1d

@leodev What’s ur verified to unverified ratio?

English

1

0

2

148

Leo - 15 y/o founder@leodev·1d

A bunch of bots are following me. I went up by 500 followers today... Elon needs to fix this bot problem, this happens to other people regularly

English

25

0

47

2.6K

Ali Romman@aliromman·1d

@ryanvogel @SpaceXAI Yeah they did, saw it in my notifications but when I checked it wasn’t there

English

0

2

1.3K

vogel@ryanvogel·1d

did @SpaceXAI get hacked?

English

15

1

154

20.5K

Ali Romman@aliromman·1d

@claudeai Anthropic is in trouble. Not because of their models intelligence, but because of their economics. x.com/aliromman/stat…

Ali Romman@aliromman

Anthropic is scared. GPT-5.6 Sol is better at coding than Fable while being half the price. Anthropic’s edge is gone. One main reason Anthropic models stayed on top is parameter-maxxing: pumping up parameters for more intelligence. The issue is rapidly diminishing returns. Opus has 5T parameters, Fable is rumored to have 10T, while 5.6 Sol is speculated at 2-4T. Larger models also use far more compute and time for both training and inference. A 10T model takes twice the compute of a 5T model. Anthropic must now choose: invest in training better models or in inference to raise limits and satisfy customers. This dilemma could have been avoided with more efficient models like OpenAI’s. Now customers are paying the price. Fable is twice as expensive as Sol, and inference costs hurt. If Anthropic removes it from heavily subsidized plans, they lose their spot in the AI race. No one will pay $10/M input for Fable when 5.6 Sol, Grok 4.5, Muse Spark 1.1, and open-weight models exist. In a nutshell, Anthropic is in trouble. Not because of their models intelligence, but because of their economics.

English

0

11

2.6K

Claude@claudeai·1d

We're extending Claude Fable 5 access on all paid plans, as well as keeping Claude Code’s weekly rate limits 50% higher, through July 19.

English

6.7K

7.3K

73.6K

26M

Ali Romman@aliromman·1d

@jyriso @nikitabier Based on how he’s taking down a bunch of spam, slop, and stolen content accounts, not really.

English

0

1

7

Jyri@jyriso·1d

@aliromman @nikitabier I bet they have more important problems to fix than this

English

1

0

10

Jyri@jyriso·2d

This is why X feels so fake. The same "Drop your SaaS link" post. 7 times in one day 😭 I don't care how well it works. I'm not building my audience like this.

English

156

3

188

12.8K

Ali Romman@aliromman·1d

@svpino That’s why it’s funny

English

0

6

235

Santiago@svpino·2d

@aliromman Sir, you are replying to a 1-yo post

English

1

0

16

1.3K

Santiago@svpino·4 Tem

you can see it, but large language models don’t. agi my ass

English

128

42

1.1K

414.5K

Ali Romman@aliromman·1d

@0xCarnagee Context rot is the bane of my existence and the biggest problem AI labs need to solve. Having more context is SO important, but it’s useless if the model loses 60 IQ whenever it’s near the cap.

English

1

0

2

80

Carnage@0xCarnagee·2d

Prime Intellect engineer: "everyone's bragging about a million-token context. here's what they don't tell you. at 256k tokens GPT-5.5 scores 80% on retrieval. push it to a million and it drops to 36%. the model accepts the context, it just can't reason across it. people call it context rot." in a 20-minute talk he explains why bigger context windows won't save your agents. continual learning + training on your own traces + real environments - that's the fix. Watch the talk, then save!

Carnage@0xCarnagee

Andrew Ng just dropped a free course on Claude Code from scratch, taught with the Anthropic team: 00:00 - why Claude Code is so agentic 04:00 - shockingly simple architecture 12:00 - point it at any codebase this short watch will replace 10 paid coding agent courses. Andrew Ng calls it his personal favorite coding assistant right now. Watch it today, then read how to engineer your own agent loops in the article below

English

91

349

4.5K

673.8K

Ali Romman@aliromman·2d

@AngelaLMorabito Since when is 30 pages large?? That should take max 45 minutes if you are a slow reader

English

0

4

212

Angela Morabito@AngelaLMorabito·2d

“professor assigned a massive PDF” 30 pages.

Português

340

482

13.9K

647.6K

Ali Romman@aliromman·2d

If you don’t read this thread at least once a week you’re ngmi

Naval@naval

How to Get Rich (without getting lucky):

English

0

5

85

Ali Romman@aliromman·2d

@birdabo Where did you hear/see this?

English

0

2

90

sui ☄️@birdabo·2d

META COOKED A SECRET MYTHOS-CLASS MODEL THEY’RE RELEASING IN A FEW MONTHS??? what.

English

30

0

128

9.4K

Ali Romman@aliromman·2d

@Polymarket Full article for those interested: pangram.com/blog/ai-in-you…

English

0

2

113

Ali Romman retweetledi

Polymarket@Polymarket·2d

NEW: Study reveals LinkedIn is the most AI-saturated major social platform, with over 40% of long-form posts flagged as fully AI-generated “slop.”

English

601

1.8K

23.1K

2.5M

Ali Romman retweetledi

Ali Romman@aliromman·5d

Maturing is realizing that NVIDIA isn’t the shovel company, it’s the company that sells the raw materials to make shovels. We dig for gold using OpenAI shovels, Anthropic shovels, and Grok shovels. Those companies in turn buy the shovel materials. It’s the great cycle of shovels that makes the world go round.

NVIDIA@nvidia

Congrats to @SpaceXAI on Grok 4.5 — trained on NVIDIA GB300 NVL72 systems and purpose-built for coding, agentic tasks, and knowledge work. This is what happens when world-class AI infrastructure meets world-class AI research.

English

0

1

6

187

Ali Romman

Keşfet