Harry Dhage

69 posts

Harry Dhage banner
Harry Dhage

Harry Dhage

@harry_dh1f

22. Founder. AI researcher Shipping AI into regulated industries (pharmacovigilance) + agentic products (https://t.co/BPFl2OebkW). Rust quant infra at night.

USA Katılım Şubat 2026
829 Takip Edilen59 Takipçiler
Sabitlenmiş Tweet
Harry Dhage
Harry Dhage@harry_dh1f·
@bindureddy If you try this with qwen modal results will be even more better and efficient.
English
1
0
0
2.1K
Harry Dhage
Harry Dhage@harry_dh1f·
Don't understand why people ask this question. Because every modal performance is best in there category.
Harry Dhage tweet media
English
0
0
0
3
Harry Dhage
Harry Dhage@harry_dh1f·
@khushiirl This is incomplete question. Because every model better in there use case
English
0
0
0
6
khushi.vy
khushi.vy@khushiirl·
be honest, which is the best open source AI model? 1. DeepSeek v4 2. Kimi K2.6 3. GLM-5 4. Qwen 3.5
English
157
6
367
39.3K
Jordan Reed
Jordan Reed@jordanreed545·
@harry_dh1f @bindureddy tuning is definitely where the magic happens! it's amazing how much it can change things. i'd love to hear more about your experience with that. let’s connect!
English
1
0
1
25
Bindu Reddy
Bindu Reddy@bindureddy·
Finally figured out how to run DeepSeek V4 Pro correctly The evals are mind blowing... Yes, it is even better than Opus 4.7 Medium!! More coming...
English
66
45
991
68.6K
Harry Dhage
Harry Dhage@harry_dh1f·
@Anaya_sharma876 Every modal has different use case for different reason. Its wrong to choose one
English
0
0
0
152
Anaya
Anaya@Anaya_sharma876·
Developers only: Which model actually writes production-ready code?
Anaya tweet mediaAnaya tweet mediaAnaya tweet mediaAnaya tweet media
English
120
23
489
51.6K
Harry Dhage
Harry Dhage@harry_dh1f·
@haider1 I think opus 4.7 works better people's don't know how to use it. Please go through the books first and understand the it's working then results will blow your mind.
English
0
0
0
23
Haider.
Haider.@haider1·
opus 4.7 is really a step back for me, so i went back to sonnet 4.6 and opus 4.5 after the first day codex 5.5 has better context retention in longer sessions, which is where claude code tends to drift on multi-file changes codex is worth giving a full month on real projects
English
22
5
152
9.7K
Harry Dhage
Harry Dhage@harry_dh1f·
@Tazerface16 It doesn't think it predict. All are prediction modals... thinking is a wrong interpretation.
English
1
0
2
166
Christopher David
Christopher David@Tazerface16·
People understand that LLMs aren't actually "thinking," right?
Drexel-Alvernon, AZ 🇺🇸 English
1.7K
705
15.7K
846K
Harry Dhage
Harry Dhage@harry_dh1f·
@trikcode Biggest mistake is it say he has completed the task when you do the gap analysis so many things left so pdf he never read fully to save a compute.
English
0
0
0
277
Wise
Wise@trikcode·
Opus 4.7 feels either confused or just risky to use. Honestly, it’s becoming the worst frontier model I’ve used in the last two years. I was hoping we’d at least get something as solid as 4.6. But 4.7 makes too many basic mistakes. You have to watch it every step. Starting to lose trust in Anthropic.
English
89
18
407
23.7K
Iván
Iván@_ivan1981_·
@nuculabs @bindureddy Front end is terrible for me. Tried to build a portfolio tracker app and it is very poor html UI. Any recommendations?
English
1
0
0
59
Harry Dhage
Harry Dhage@harry_dh1f·
@bindureddy If you try this with qwen modal results will be even more better and efficient.
English
1
0
0
2.1K
Harry Dhage
Harry Dhage@harry_dh1f·
What’s something people chase that isn’t worth it?
English
0
0
0
27
Harry Dhage
Harry Dhage@harry_dh1f·
@OrevaZSN Absolutely true.. but x create a lots of values and people makes good amount money too to support there project.
English
0
0
0
22
Harry Dhage
Harry Dhage@harry_dh1f·
@APompliano This before mythos think about after persistent modals. Expected guess??
English
1
0
1
21
Anthony Pompliano 🌪
Anthony Pompliano 🌪@APompliano·
Anthropic is now reportedly doing $44 billion in annual revenue. That means they added $14 billion in the last 30 days. Almost half a billion dollars in revenue PER DAY. Insane.
English
202
108
2.1K
158.3K
Harry Dhage
Harry Dhage@harry_dh1f·
I spent 1.7B tokens on Codex (Pro 5x) and 80M tokens on Claude Code (Max 20x) yesterday. Guess which one gave me a usage-limit warning.
English
0
0
0
63
Peter H. Diamandis, MD
Peter H. Diamandis, MD@PeterDiamandis·
For those worried about any coding language... Your 10 years mastering Python is about to become a commodity. But your ability to think, lead, and create is not.
English
140
97
1.1K
43.2K
Harry Dhage
Harry Dhage@harry_dh1f·
Started reading this book!! I will be adding important points evey day about this. So stay tuned for the updated and increase you knowledge.
Harry Dhage tweet media
English
0
0
2
35
Harry Dhage
Harry Dhage@harry_dh1f·
@realEstateTrent That's the reason before hiring we should not ask experience. If you are not building a tools. Where research is needed.
English
0
0
1
726
StripMallGuy
StripMallGuy@realEstateTrent·
We hired a 27-year-old associate a few months ago, and he uses Claude Cowork all day. I don't have the words to describe what is happening to our productivity. Things that would have taken months are now getting done in hours - it's changing the entire way we run our business!
English
251
148
6.9K
776.8K
Harry Dhage
Harry Dhage@harry_dh1f·
@sama Where feel codex should be used a lot if we wants to best out of it?
English
0
0
0
18
Sam Altman
Sam Altman@sama·
you know what all of these "which is better" polls are silly use codex or claude code, whatever works best for you i am grateful we live in a time with such amazing tools, and grateful there is a choice
English
2.2K
1.1K
23K
1.6M
Harry Dhage
Harry Dhage@harry_dh1f·
@MrBeast @DOUZIE25 I'm remembering facebook scam that is to come in early of facebook like click here nad there
English
0
0
0
15
MrBeast
MrBeast@MrBeast·
If this tweet has exactly 1 like in 24 hours I’ll give that person $1,000,000
English
264.5K
31.1K
1.1M
69.4M